Carnegie Mellon University
Below you will find a compilation of the graduate participants who exhibited at Innovation with Impact in 2017 listed alphabetically by school, department, and last name. Included are the participant's name, exhibit title, and abstract.

Biomedical Engineering

Stephanie Beels, Pritham Aravind , Suyash Kela, Raafe Khan, Apratim Vidyarthi
Saudi Arabia in the Sun
This study has determined that completely phasing out oil from the electricity mix in KSA is not  recommended, but rather that KSA should adopt a mixed energy portfolio with a total installed  capacity of 120 GW by 2032, with 34% solar photovoltaic (PV) energy, 41% natural gas,  10% oil, 8% nuclear energy, and 7% wind and geothermal energy. Thus, a minimum percentage of the electricity mix must include 10% electricity generated from crude oil. This is because renewable energy has not and will not have matured enough in terms of efficiency and  equity to make it a reliable and feasible source of continuous energy. Natural gas and nuclear power plants should be secured since they are the most reliable sources of energy.  The 41 GW of solar capacity projected to be installed by 2032 will be split up into three potential  solar parks, placing 10 GW each near Mecca and Riyadh and the remaining 21 GW in the southern  desert region near Bisha. The area of the parks around Mecca and Riyadh are 14x14km each, and  the area of the park around Bisha is 17x17km. This siting strategy will ensure minimum electricity transmission costs and maximum reliability & capacity utilization.  According to the cost-benefit analysis, this project is extremely profitable over the stipulated time  period and can successfully be self-funded. The private net present value (NPV) is estimated to be around $220 billion, with the social NPV estimated at $502 billion. This analysis did not consider miscellaneous costs, uncertain policy costs and purchasing costs, but given that the calculated NPV is sufficiently high, these costs can confidently be compensated to make the project break-even over its lifetime.  The project, although seemingly very profitable, has various risks associated with it. Most of these  risks are either low or medium risks, with the most challenging ones being the political unrest  surrounding the Middle East which could possibly disrupt the timeline of the project. Another major risk is the environmental/social risks associated with the project, which could unsettle the natural ecosystem and can further aggravate the already environmentally-stressed nation. Both these issues can be effectively mitigated with awareness and defensive preparations, which could be out of scope for this project.  On a geopolitical forefront, the major objective of the KSA government is to ensure its energy security, which can be done by carefully siting and constructing the solar capacity in a carefully selected, secure location. Also, with a project of this magnitude, KSA must be cognizant of the  effects it can have on their international, political allies and enemies. This project could also have a large influence on KSA’s usage of natural gas for electricity production. The installation of alternative energy could motivate the government to phase out non-renewable energies in the long term, after 2032.

Alan Campbell                                        
Polymer-Based Protein Engineering for the Modification of Redox Enzyme Activity, Stability and Electron Transfer Efficiency    
The implementation of enzymatic biosensors and biofuel cells is currently limited by two major factors: low current density and low stability/reliability. Solving these issues would allow for the development of medical and commercial technologies that could improve the lives of millions. In our preliminary work we identified the major sources of electron transfer limitation and operational/storage stability in a high performing system. We now propose research aimed to enhance the capabilities of electroactive enzyme-based systems through the utilization of polymer-based protein engineering by developing an increased understanding of these limitations and their causes as well as the interactions between polymer-modified electroactive enzymes and their operational environment.                  

Guruprasad Raghavan, Qinle Ba                                        
Computational and Experimental Analysis of Organelle Networks    
Membrane-bound organelles are biochemically distinct compartments used by eukaryotic cells for serving specialized physiological functions and for organizing their internal environment. Recent studies have revealed extensive communication between both homotypic organelles and heterotypic organelles through mechanisms such as membrane contacts and vesicle fission and fusion. Currently, however, how the organelle networks are spatially organized and interact at the whole-cell scale remains poorly understood. In particular, it remains unknown whether, and if so how, interactions between different organelle networks may affect their spatial distributions, and vice versa. To address these questions, we developed image-based computational analysis methods and used them to examine the spatial distributions of the lysosome network and the peroxisome network and their interactions in cultured COS-7 cells. Because variations in cell shapes may influence spatial distributions of the organelles, we cultured the cells on printed circular protein substrates so that they all had approximately the same shape and size. We first examined the global distribution patterns of the lysosome network and the peroxisome network by developing four important spatial statistical metrics. We then investigated their interactions locally by following individual lysosomes and peroxisomes using single particle tracking. We characterized relations between their distributions using the Bhattacharyya distance between probability distributions, and Hausdorff distance between probability density maps. To validate the use of metrics such as inter-organelle distribution, nearest neighbor distribution, we treated the cells with tBHP, an agent that induces oxidative stress in the cell, and forces redistribution of lysosomes to the perinuclear region. We found evidence of spatiotemporal coordination within organelle networks under the applied perturbation. Together, our data provide new insights into how organelle networks are spatially organized and how their interactions modulate their distributions globally at the whole-cell scale.        

Elaine Soohoo                                        
Computational Assessment of Design Parameters for a Torsional Ventricular Assist Device (tVAD)    
The goal of this study is to quantify changes in hemodynamic function with the application of apical torsion as a means of cardiac assist. Parametric computational simulations utilizing a beating heart model of both ventricles attached to a closed-loop circulatory system was used to determine design parameters for a second-generation tVAD prototype. Varying degrees of applied apical torsion were simulated by altering device coverage areas and rotation angles to compare the effects on global cardiac function to a clinical pretreatment, No-VAD model.  Increasing these parameters produced increases in ejection fraction, peak systolic pressure, and stroke work while lowering end-systolic volume.             

Chemical Engineering

Rebecca Ball                                        
Lipidoid siRNA Nanoparticles for Inflammatory Bowel Disease Therapeutics    
Inflammatory bowel disease (IBD) is an intestinal malady that is associated with damaging symptoms, frequent relapses, and treatments that are hindered due to the lack of knowledge of the underlying physiological cause of the disease. Here, we utilize nanoparticles made of lipid-like molecules, which we call lipidoid nanoparticles (LNPs), to deliver short interfering ribonucleic acid (siRNA) for intestinal disease therapeutics. siRNA therapeutics show great promise for the treatment of intestinal diseases such as IBD and gastrointestinal cancer due to siRNA’s ability to specifically suppress the expression of a protein of interest.  

Toni Bechtel                                        
Nonlinear Relaxation Modulus via Medium Amplitude Oscillatory Shear (MAOS)    
We develop a general framework for determining the relaxation moduli of a material from a medium amplitude oscillatory shear (MAOS) deformation. Knowledge of the relaxation moduli of a given material allows for one to predict the stress response of the material under any arbitrary transient deformation via a memory integral expansion. This framework is demonstrated by deter- mining the first nonlinear relaxation modulus, referred to as the MAOS relaxation modulus, for a dilute suspension of Brownian spheroids in a Newtonian fluid subject to a dual-frequency MAOS deformation. We first determine the general expression for the second normal stress difference from the memory integral expansion. Second, the microstructural stress response of the model is system is determined via a regular perturbation expansion of the Fokker-Planck equation at small dimensionless strain-rate amplitude, or Weissenberg number (Wi). An analytical expression for the MAOS relaxation modulus is resolved by comparing the second normal stress difference result of the memory integral expansion and microstructural stress response. Finally, we reconstruct the stress response of a dilute suspension of both rods and disks for the start-up and cessation of steady shear and steady uniaxial extension.                

Nicholas Lamson                                        
pH-Predicted Behavior of Piperazine Derivatives as Transepithelial Permeation Enhancers    
Effective oral delivery of macromolecule drugs (e.g. insulin for diabetics) has long eluded the field of biomedicine. This work aims to break down one of the major barriers at play by examining molecules that act on the intestines to increase drug transport into the bloodstream. Fourteen members of the piperazine chemical family are shown to boost bioavailability of model drugs through a simulated intestinal lining. Interestingly, the most prominent structure-function relationship identified dictates that nontoxic but effective treatments fall within a narrow pH window of 8.7 to 9.6.

Rajarshi Sengupta                                        
The Role of Surface Charge convection in the Electrohydrodynamics and Breakup of Fluid Drops    
The deformation of a weakly conducting leaky dielectric drop in a density matched, immiscible, weakly conducting medium under an external electric field is quantified computationally. A tangential electric stress acts along the interface, which drives both fluids into motion, causing convection of surface charge carriers along the interface. This effect is measured in terms of an electric Reynolds number. Using boundary integral computations, we show that the electric Reynolds number can change the breakup mode of a weakly conducting prolate drop from end-pinching to tip-streaming, if the induced flow is directed from the equator of the drop to its poles.

Charles Sharkey                                        
Effect of Particle Hydrophobicity on the Shapes of Non-Spherical Capsules    
We stabilize air bubbles by adsorbing surfactant coated silica nanoparticles to the air-water interface of a bubble in a confined geometry. The surfactant molecules vary the particle hydrophobicity. The result is non-spherical stabilized bubbles, with shapes ranging from spherical to nearly cylindrical. We vary surfactant concentration to evaluate the role of hydrophobicity in determining capsule shape. Bubble speed is also controlled. Bubbles produced in suspensions with higher surfactant concentrations are more cylindrical in shape. Additionally, capsules are less cylindrical as the speed of the bubble decreases.  We identify several mechanisms to explain observed differences with previously predicted trends.   

Civil & Environmental Engineering

Xiaoju Chen                                        
Uncertainties of Energy Consumption for Industry Based on a Matrix-based Life Cycle Assessment Model    
Large uncertainties exist in the energy consumption across all industries and their supply chains. Previous research has estimated the uncertainties due to specific products or processes through case studies, however, systematic uncertainty estimation on all of the existing US industry sectors has not been addressed. We provide a new method to determine the uncertainties of energy consumption and environmental impacts of all sectors as well as their supply chains. The method applies a wide range of publicly available inventory data to evaluate bounding results as energy uncertainties for each industry. Economic Input-Output Life Cycle Assessment model is used to demonstrate the method.

Kyle Gorkowski                                        
Air Pollution Studies with the Aerosol Optical Tweezers    
Atmospheric aerosols contain a wide variety of organic and/or inorganic components and can phase-separate into distinct liquid phases, resulting in either a core-shell or a partial-shell particle morphology. Understanding and predicting when each of these morphologies forms is critical to understanding gas-particle interactions and climate-forcing properties. We conducted experiments exploring phase-separation of droplets suspended using aerosol optical tweezers (AOT). The droplet levitation and the surface resonant whispering gallery modes (WGMs), retrieved in the cavity enhanced Raman spectrum, provide a unique direct and real-time assessment of the droplet’s morphology.

Yiming Gu                                        
Bayesian-based Traffic State Estimation    
Traffic state estimation (TSE) aims to estimate the time-varying traffic charac- teristics (such as flow rate, flow speed, flow density, and occurrence of incidents) of all roads in traffic networks, provided with limited observations in sparse time and locations. TSE is critical to transportation planning, operation and infrastruc- ture design. In this new era of “big data”, massive volumes of sensing data from a variety of source (such as cell phones, GPS, probe vehicles, and inductive loops, etc.) enable TSE in an efficient, timely and accurate manner. This research develops a Bayesian-based theoretical framework, along with statistical inference algorithms, to (1) capture the complex flow patterns in the urban traffic network consisting both highways and arterials; (2) incorporate heterogeneous data sources into the process of TSE; (3) enable both estimation and perdition of traffic states; and (4) demon- strate the scalability to large-scale urban traffic networks. To achieve those goals, a Hierarchical Bayesian probabilistic model is proposed to capture spatio-temporal traffic states. The propagation of traffic states are encapsulated through mesoscopic network flow models (namely the Link Queue Model) and equilibrated fundamental diagrams. Traffic states in the Hierarchical Bayesian model are inferred using the Expectation-Maximization Extended Kalman Filter (EM-EKF). To better estimate and predict states, infrastructure supply is also estimated as part of the TSE pro- cess. It is done by adopting a series of algorithms to translate Twitter data into traf- fic incident information. Finally, the proposed EM-EKF algorithm is implemented and examined on the road networks in Washington DC. The results show that the proposed methods can handle large-scale traffic state estimation, while achieving superior results comparing to traditional temporal and spatial smoothing methods.

Joe Moore                                        
Differential Effects of Copper Oxide Nanoparticles and Cu2+ on Bacteria    
We explored effects of and responses to an increasingly popular antimicrobial engineered nanomaterial (ENM), copper oxide (CuO), in Escherichia coli and laboratory and clinical isolates of Staphylococcus aureus, including methicillin-resistant strains. We found Escherichia coli to be significantly more sensitive to CuO ENM exposure under anoxic conditions. RT-qPCR showed that, under oxic conditions, CuO ENMs induced a similar transcriptional response to that of Cu2+. Hyperspectral imaging and RT-qPCR results showed CuO ENM affinity for bacteria. Our results could help ensure that antimicrobial ENMs developed for water treatment processes and other applications appropriately balance effectiveness and potential for downstream environmental impact.


Electrical and Computer Engineering

Xinlei Chen                                        
HAP: Fine-Grained Dynamic Air Pollution Map    
This paper presents a hybrid adaptive particle filter (HAP) with online feedback to dynamically reconstruct high spatial-temporal resolution air pollution information from sparse vehicular based sensors. To deal with data sparsity, we apply both spatial and temporal correlation of air dispersion to reduce data dimension requirement. HAP adaptively predicts when the accumulated prediction error is low and then uses data compensation for correction whenever the prediction error becomes high. The preliminary results based on the city scale deployments with 10 taxis show that our system achieves up to 50% reduction on system errors.

Amit Datta                                        
Evaluating Privacy and Fairness in Data-driven Decision Systems    
Data-driven systems are increasingly being employed to make critical decisions about our lives. Such systems utilize the power of big data to make ever so accurate predictions. The widespread use of such systems has led to concerns over societal values like user privacy and fairness. Although these systems have policies promising to protect societal values, the blackbox nature of these systems makes it difficult to analyze whether these policies are respected. Moreover, it is difficult to hold entities accountable upon detection of a policy violation. We show that it is possible to evaluate some privacy and fairness properties on fully blackbox decision-making systems. In particular, we study discrimination, transparency and choice on Google's ad ecosystem. We also present some initial approaches to enable accountability for fairness violations in systems with access to the internal structure and logs. By enabling accountability, the objective to assign responsibility for violations to internal modules of or inputs to the system. We develop our theory of accountability in collaboration with Microsoft Research with Bing's advertising pipeline as a possible application.   

Sanghamitra Dutta                                        
Short Dot: Computing Large Linear Transforms    
Faced with saturation of Moore’s law and increasing size and dimension of data, system designers have increasingly resorted to parallel and distributed computing to reduce computation time of machine-learning algorithms. However, distributed computing is often bottle necked by a small fraction of slow processors called “stragglers” that reduce the speed of computation because the fusion node has to wait for all processors to complete their processing. To combat the effect of stragglers, recent literature proposes introducing redundancy in computations across processors, e.g., using repetition-based strategies or erasure codes. The fusion node can exploit this redundancy by completing the computation using outputs from only a subset of the processors, ignoring the stragglers. In this paper, we propose a novel technique – that we call “Short-Dot” – to introduce redundant computations in a coding theory inspired fashion, for computing linear transforms of long vectors. Instead of computing long dot products as required in the original linear transform, we construct a larger number of redundant and short dot products that can be computed more efficiently at individual processors. Further, only a subset of these short dot products are required at the fusion node to finish the computation successfully. We demonstrate through probabilistic analysis as well as experiments on computing clusters that Short-Dot offers significant speed-up compared to existing techniques. We also derive trade-offs between the length of the dot-products and the resilience to stragglers (number of processors required to finish), for any such strategy and compare it to that achieved by our strategy.

Benjamin Elizalde                                        
An Approach for Self-Training Audio Event Detectors    
Audio Event Detection (AED) aims to recognize sounds within audio and video recordings and employs machine learning algorithms trained and tested on annotated datasets. However, available datasets are limited in number of samples and hence it is difficult to model the acoustic diversity. Therefore, we propose combining labeled audio from a dataset and unlabeled audio from YouTube to improve the sound models. The performance of the re-trained detectors are compared to the original detectors using the annotated test set. Results show improvement of AED after re-training as well as uncovering challenges of using web audio from videos.

Ece Ozalp, Sungho Kim                                        
Fabrication of Conductive Nanoporous Membranes    
This paper reports on the fabrication of an intrinsically  conductive nanoporous membrane with high pore density (10^11 pores per cm^2) using self-assembling block copolymers which has the potential to enable the controlled diffusion of charged molecules via the exclusion-enrichment effect, with potential applications in drug-delivery. Finite element simulations of an electrically gated nanochannel indicate the possibility to alter the diffusion rate of molecules through the membrane by several orders of magnitude using a ±1 gate voltage. The nanochannels with 25 nm diameter pore size were fabricated into an amorphous silicon thin film sheathed by native oxide in order to enable gating. Poly (styrene-bdimethyl  siloxane) block copolymer was first used to  generate a highly dense array of pillars; the pattern was then reversed and subsequently transferred into the amorphous silicon thin film.

Viswa Tej Koganti                                        
Applications of Eye-Gaze Tracking to Vehicular Path-Prediction    
Using a Vehicular Wireless Network to send basic safety messages (BSMs) containing information such as speed and direction is an important safety application of wireless technology in automobiles. This project studied the potential for using computer vision to augment the information provided by these BSMs. The hypothesis was that determining which part of the road the driver is looking at can be used to predict the path of their vehicle. In live-testing, a minimum of 70% accuracy for time-lagged eye-gaze tracking matching vehicular motion was achieved. This is significant and useful for advisory parameters to current path prediction algorithms.

Swadhin Thakkar, Krishna Kumar, Sharath Rangan            
Marauder's Map    
A locating based app to connect people on CMU campus. It uses indoor location tracking technology for peer-to-peer location sharing and provides to platform to connect various communities on campus.

Diyi Yang                     
Identifying Semantic Edit Intentions from Revisions in Wikipedia    
Extracting the intention behind changes to documents can bring deep insights to collaborative writing processes. In this work, we developed in collaboration with Wikipedia editors a 13-category taxonomy of the semantic intention behind edits to Wikipedia articles.  Using labeled article edits, we build a computational model that achieved a micro-averaged F1 score of 0.621 across edit intentions. We then used this model to investigate how different types of edits predict the retention of newcomers and the changes in the quality of articles, two key concerns for English Wikipedia today. Our analysis shows that the types of edits that users make in their first session predict their subsequent survival as Wikipedia editors. In addition, edit intention predicts improvements in article quality, and articles seem to need different types of edits depending upon their initial quality.

               

Engineering & Public Policy

Matthew Babcock                                        
Exploring How a Boundary Organization Network Increases Rancher "Buy-in" on Climate Issues in Montana    
Research on communicating climate and adaptation information between stakeholder groups has shown the importance of boundary organizations, especially in areas with existing societal divisions. We conducted in-depth mental model interviews with members connected to an organization trying to bridge the urban/rural and scientists/agriculturalist divisions on climate in Montana. We use this qualitative data, and some limited network and survey data, to explore how the boundary organization network came to be and what strategies are successful in communicating climate adaptation information with local ranchers. We compare our findings with that of the existing research on climate communication and boundary chains.
        
Barry Dewitt                                        
Heterogeneity in Preferences for Health    
Health-related quality of life (HRQL) is used extensively to quantify the effectiveness of medical interventions. Societal preference-based HRQL measures aim to produce societal utilities for health by aggregating valuations from individuals in the population. This preference aggregation is governed by social choice theory, a normative theory with implications for the construction of societal preference-based HRQL measures. Here, we present preliminary analyses of preference heterogeneity for different types of health. Our goals are to characterize this heterogeneity, and then use that characterization to provide policy recommendations for incorporating it into models of health utility, in a way that is normatively grounded in social choice theory and reflects the best practices of behavioral decision research.

Daniel Gingerich                                        
Life-Cycle Air Emission Damages for Municipal Drinking Water Treatment    
Risk tradeoffs currently considered for drinking water treatment and regulations only include those risks associated with the treated water and to consumers.  In this project, we incorporate life-cycle thinking into risk tradeoff analyses for drinking water to expand the exploration of risk tradeoffs to include risks from air pollution and populations that do not directly consume drinking water.  We calculate the life-cycle air emissions and damages from electricity and chemicals consumed during the treatment process for currently installed treatment processes and six different regulations.  We then compare these damages from drinking water benefits to the benefits of these interventions.

Brock Glasgo                                        
Assessing the Value of Information in Residential Building Simulation    
Building energy simulation models are now being used in a broad range of applications including building and retrofit design, building operation optimization, performance verification for energy efficiency programs, and – recently – national energy code analysis and design. The growing role that these software tools play in informing policy and investment decision-making has created a need to verify the accuracy of their results and develop methods for calibrating them to ensure reliable outputs. We compare the results of simulations to device-level monitored data from actual homes to provide a first measure of the accuracy of the EnergyPlus device-level estimates in a large number of homes. We then conduct sensitivity analysis to observe which physical and behavioral characteristics of the homes and homeowners most influence the accuracy of the modeling.

John Helveston                                        
Innovating Up, Down, and Sideways: The (Unlikely) Institutional Origins of Experimentation in China's Plug-in Electric Vehicle Industry    
A vast literature has attempted to understand the factors that accelerate experimentation and innovation in technologically-sophisticated emerging industries—but less is known about these processes in the context of industrializing nations. We apply inductive, grounded theory-building techniques to characterize and explore the origins of divergent innovation trajectories in once such context: the plug-in electric vehicle (PEV) industry in China. Triangulating annual vehicle make and model sales data for 2003-2014 (plus monthly data for the most recent five years); 112 English and Mandarin archival documents from industry, academic, and news outlets; and 51 semi-structured interviews across industry, government, and academic stakeholders, we develop four in-depth case studies. We find that in contrast to the innovation trajectories of the multinational and Chinese arms of joint venture (JV) firms, independent domestic Chinese firms (those with no historic JV partnerships) are undertaking significant innovation and experimentation in China’s PEV industry. Our results suggest that national institutions—specifically the formal JV and local content requirements—which discouraged PEV innovation in multinational firms and inhibited the capabilities of Chinese JV partners to independently develop their own PEVs resulted in a protected PEV market for independent domestic firms. The influence of these national institutions has combined with local institutional support in the form of additional market protection and subsidies to turn regional markets into protected laboratories for independent domestic firms to experiment with a variety of innovations. That said, for these domestic firms to grow beyond their early, protected regional markets, China will need to develop paths to national market integration.

Daniel Sun, Timothy Bartholomew, Paul Welle, Sneha Shanbhag, Daniel Gingerich                
Effect of Coal Quality and Air Pollution Control Devices on Trace Element Emissions at Coal-fired Power Plants    
We estimate the partitioning of mercury, selenium, arsenic, and chloride in the bottom ash, fly ash, and chloride purge, gypsum, and stacks for each coal fired power plant using power plant characteristics and trace element concentration of coal combined with a mass balance model. Our estimate of Hg emissions in the stacks against monitored Hg emissions data from 98 power plants show that we overestimate Hg emissions by at least a factor of two for 60% of boilers while predicting 21% of boilers within a 50% margin of error. Despite inaccuracies, the model can estimate emissions relative to plants, which is useful as a cheap alternative for monitoring and assessing risks from trace elements.

Fan Tong                                        
A Spatial Assessment of Climate Change Damages and Air Pollution Damages From Light-duty and Heavy-duty Vehicles in the United States    
This paper estimates the life cycle climate change damages and air pollution damages of five fuel pathways in five types of light-duty and heavy-duty vehicles used in each county across the United States. It shows that social cost of carbon and value of a statistical life have large impacts on the ranks of fuel-vehicle choices considered. Our results show that considering only climate change damages or only air pollution damages lead to similar fuel-vehicle choices for light-duty vehicles but different choices for heavy-duty vehicles, thus highlighting the trade-offs between climate change mitigation and air pollution mitigation in the transportation sector.  

Engineering & Technology Innovation Management

Apratim Vidyarthi                                        
CGIU: Exploring the Potential for Developing a Refurbished Laptop Supply Chain in India    
Our original goal was to implement a refurbished-laptop supply chain for impoverished children in India, by recycling corporate laptops. However, we had several roadblocks that showed that because of archaic policies, this was impossible to implement.

Energy Science, Technology & Policy

Yoolhee Kim                                        
Potassium Oxygen Battery    
I will either present a poster or submit a short paper on what I gained from the weSTEM conference which I was able to attend through the GSA conference funding. The poster would be on my ongoing research in potassium oxygen batteries, and using DFT to determine various material properties of the discharge product.

Information Networking Institute

Vrushali Bhutada                                        
Grace Hopper Conference    
I attend Grace Hopper Conference for networking and career fair.

Manideep Konakandla                                        
Security Research on Docker Containers    
The concept of containerization was in Linux from ages in the form of jails, zones, LXC etc. but it is since 2 years it gained tremendous recognition. The credit goes to "Docker" which made the concept of containerization very useful and handy by adding many benefits to existing container technologies. Tech giants like Redhat, Google, IBM, VMware etc. are not only the biggest contributors to this most active open source project but also major users of it. The effect of containers already impacted the virtual machine market and this impact is going to increase significantly in the near future.    Security is always an important issue for any upcoming technology and Docker is no exception to it. This presentation starts with a brief introduction to containers vs. virtualization technology, Docker ecosystem and then goes in-detailed into “Docker Security”. It gives you an overview of security issues that can occur at every point in Docker container pipeline but goes “deep” into security issues of “Images” and “Container run-time”. Then, you will be learning on how to protect your container ecosystems from these security issues. Presentation also covers big-list of enterprise specific  container security measures, golden rules to maintain each component of your container ecosystems securely,  building a secure in-house Docker images registry, creating enterprise level container security standards and guidelines, Tools for your container ecosystem, hardware isolation to containers etc. By the end of the presentation, you will be able to assess how secure your container pipeline is and take defensive measures accordingly.

Elomar Souza                                        
RailsConf 2016: Rails as Mature Platform for New Products    
Out of the spotlight as a new, cutting-edge edge tool for startups, Rails is now in a mature and stable stage as a platform for new products. RailsConf 2016 discussed this stage and what's next for Rails and web development, here reviewed through four topics: improvements for the upcoming needs of web apps, optimization work, industry maturity, and what comes next.

Materials Science & Engineering

Yu-Han Liang                                        
Liquid-metal-enabled Synthesis of High-aluminum-containing III-nitrides by Plasma-assisted Molecular Beam Epitaxy for Device Applications    
We have studied MBE-grown quantum well structures and high-aluminum-containing (Al,Ga)N films doped with Mg for potential LED applications. These structures and films were grown under the liquid-metal-enabled growth mode.

Ankita Mangal                                        
Applied Machine Learning to Predict Stress Hotspots in Materials    
Stress hotspots are regions of high stresses in a microstructure, and they also tend to have large stress gradients. They are correlated with the nucleation of damage such as voids in ductile fracture. Stress hotspots tend to form near microstructural features, such as grain boundaries, triple and quadruple junctions and usually form in textures corresponding to maxima in the Taylor factor for a given loading condition. We simulate deformation in a uniaxial tensile test in cubic polycrystals with equiaxed grain structures, using the elasto-viscoplastic Fast Fourier Transform (EVPFFT) code. After identifying stress hotspots by thresholding, we characterize their neighborhoods using various metrics that reflect local crystallography, geometry, and connectivity. This data is used to create input feature vectors to train a supervised machine-learning algorithm, which predicts the location of stress hotspots. The results show the power and the limitations of the machine learning approach applied to the polycrystalline grain network.

Mechanical Engineering 

Zack Francis                                        
Spot Size Adjustments to Reduce Flaws and Expand Processing Space    
Direct metal additive manufacturing is a complex process that builds up parts by focusing a heat source to create melt pools on the substrate or existing deposit. The spot size, or beam diameter, of the heat source directly impacts the melt pool geometry, porosity, and other deposition properties. A simplified FEA was developed to simulate various spot sizes across different processes and alloys. Changes in melt pool geometries for different spot sizes are related across multiple processes and alloy systems. A method for avoiding keyholing and expanding the usable processing space is presented.

Kosa Goucher- Lambert, Chris McComb                                    
Impossible by Design? Fairness, Strategy, and Arrow's Impossibility Theorem    
The design process often requires work by teams, rather than individuals. During team-based design it is likely that situations will arise in which individual members of the team have different opinions, yet a group decision must still be made. Unfortunately, Arrow’s Impossibility Theorem indicates that there is no method for aggregating group preferences that will always satisfy a small number of “fair” conditions. This work seeks to identify methods of combining individual preferences that can come close to satisfying Arrow’s conditions.

Hugh Li                                        
Spatial Variation of Organic Aerosol and Source Identification of Temperature-resolved Carbon Fractions    
Long-term exposure to particulate matter (PM) is the major contributor to air pollution related death in 21st century. Organic aerosol is a major component of PM2.5, ranging from 20% to 90% in mass. A mobile sampling campaign was conducted in 2013 summer and winter in Pittsburgh, PA and to characterize spatial variations in organic aerosol mass and its components. 36 sites were chosen based on three stratification variables—traffic density, proximity to point source, and elevation. Filter samples were collected in three time sessions (morning, afternoon, and night) in each season.

Nathan Nakamura                                        
Low Temperature Direct Synthesis of Mixed Amorphous-Crystalline TiO2 Thin Films Under Electromagnetic Excitation    
We utilize synchrotron x-ray radiation and thin film pair distribution function (tfPDF) analysis to show that EM fields, specifically microwave radiation (MWR), enables single-step synthesis of mixed amorphous-crystalline titanium dioxide (TiO2) films and achieves crystallinity at reduced reaction temperatures and times. More importantly, our experiments demonstrate that films grown by MWR-assisted synthesis contain different and more crystalline phase composition compared to TiO2 grown at similar temperatures without EM field exposure. MWR-grown films are composed of a mixed-phase structure consisting of long-range anatase TiO2 with short-range amorphous components, while furnace-grown materials are amorphous with local ordering most resembling the brookite phase.

Vikram Pande                                        
Design Principles for Electrolytes in Li-O2 batteries    
Among the “beyond Li-ion” batteries, non-aqueous Li-O2 batteries have the highest theoretical specific energy  and, as a result, have attracted significant research attention over the past decade. A critical challenge facing non-aqueous Li-O2 batteries is low cell capacities. Recently, it was shown that, to improve capacity, we need to enhance solubility of the intermediate LiO2*. This can be done by either solvating the Li+ cation or the superoxide anion O2-. However, increasing solvation leads to lower conductivity, rechargeability of cells. In this study, we report fourfold enhancement in Li-O2 cell capacity by appropriately selecting the salt-solvent combination in the electrolyte.

Sneha Prabha Barra                                        
Spatial Control of AM Solidification Microstructure Across Multiple Alloys and Processes    
In this work, previous results for solidification microstructure control in Ti-6Al-4V for single bead and thin wall geometries are extended to solid build geometries for multiple alloy systems and two processes. Location- specific microstructure control via additive manufacturing has been demonstrated by the authors for Arcam deposits of Ti64 but has not been explored in detail.  Also, it is important to understand relationship between process variables and microstructure while minimizing part porosity, optimizing the build rate or controlling other process outcomes. Toward this goal, solid test blocks of Ti-6Al-4V, IN718 and AlSi10Mg are built by varying process variables.  Results show a fundamental relationship between melt pool geometry and solidification microstructure across different alloys and different processes. The result is applied to achieve location specific microstructure control in a target component.   

Avinash Siravuru                                        
Optimal Control for Geometric Motion Planning of a Robot Diver    
Inertial reorientation of articulated bodies has been an active area of research in the robotics community as this behavior can help guide dynamic robots to a safe landing with minimal damage.  The main objective of this work is emulating the aggressive and large angle correction maneuvers, like somersaults, that are performed by human divers. To this end, a planar three link robot, called DiverBot, is proposed.  By considering a gravity-free scenario, a local connection is obtained between joint angles and the body orientation, resulting in a reduction in the system dynamics.  An optimal control policy applied on this reduced configuration space yielded diving maneuvers that appear human-like  while being dynamically feasible.  Numerical results show that the DiverBot can execute one somersault without drift and multiple somersaults with drift.

Phil Smith                                        
Ultra-thin Silicon Solar Cells by Large-scale Electronic and Optical Interface Engineering    
In silicon solar cells, surface texturing and passivation techniques are incorporated into the fabrication process to minimize optical and charged carrier recombination losses. In this project the use of nanosphere lithography and poly(3,4 ethylenedioxythiophene) (PEDOT) will be used as two cost effective means for texturing and passivating silicon solar cells respectively. The PEDOT films are deposited via chemical vapor deposition which provides an inexpensive, solvent free route towards the deposition of this conducting polymer. Nanosphere lithography is a “bench top” technique relying on the self-assembly of a monolayer of nanospheres which forms the basis for creating textured surfaces.

Mark Whiting                                        
Automated Induction of General Grammars for Design    
Grammars are useful for representing systems of design patterns, however formulating grammars is not straightforward due to challenges of representation and scope. This work introduces a highly flexible mechanism for producing viable grammars from a computational representation of a designed context. This mechanism is evaluated with a range of common types of devised media of increasing complexity based on dimensionality and, against a set of grammar properties necessary for a grammar induction to be useful: accuracy, variability, repeatability and conciseness. Arithmetic completeness in the 1D case and a continuum of approaches for higher dimension data sets are shown.

Hesham Zaini, Ayush Agrawal                                        
Experimental Gait Analysis of Waveboard Locomotion    
Through modeling and experimentation, we analyze common gaits on a waveboard, an underactuated mechanical system whose motion is governed by both nonholonomic constraints and momentum conservation. We take advantage of the system's symmetries to derive a reduced system model that differentiates between kinematic and dynamic components of motion, which we evaluate using marker trajectory data gathered through an optical tracking system for various types of gaits. By extracting relevant parameters, we determine the kinematic components of gaits commonly used by human riders. In particular, we demonstrate that traditional forward motion is purely dynamic, while sustained turning motion contains kinematic components.






 

Architecture

Javier Argota Sanchez, Camille Baumann-Jaeger, Scott Donaldson, Cecilia Ferrando
MS in Computational Design in Process    
As part of our research within the MS in Computational Design, we will be presenting insights and evolution of our inquiries in relation to some projects presented in the ACADIA 2016 Posthuman Frontiers congress.

Camille Baumann-Jaeger, Javier Argota, Scott Donaldson, Cecilia Ferrando                    
ACADIA and Computational Design    
Last year, my classmates and I attended ACADIA, a conference that focuses on the intersection between technology, design, and architecture.  

Art  

Shobun Baile                                        
Antisuicide Design    
My work uses renderings of 3D models of architectural elements made to be resistant to suicide to explore the relationship between minimalism as a principle of aesthetics and as a necessity of function.

Alex Lukas                                        
Giant Concrete Arrows Dot the Landscape as Relics of Early Transcontinental Airmail Service    
Starting in 1924 the first transcontinental airmail routes across the United States were marked by a series of 60 foot long concrete arrows placed on the ground every ten miles to direct pilots to their destination. While many of these arrows no longer exist, some remain. In August of 2016, with the support of a Carnegie Mellon University GuSH grant, I traveled to visit seven of these concrete arrows in Southern California, Nevada and Utah, and visited historical collections focused on early aeronautics in archival repositories at the University of Southern California and the Los Angeles Natural History Museum.

Daniel Pillis                                        
Virtual Newell Simon Simulation    
A simulation of the founding of artificial intelligence.

Katie Rose Pipkin                                        
Xray Mutanegenesis as a Generative Practice    
I will be presenting artworks that circle in and around generative practices, including software bots, generated books, and experiments and research in plant mutation.


Design

Michael Arnold Mages                                        
Uber and Language-Action Theory    
Mediated communication is the way that distributed and proximate work teams communicate, and is structured nearly completely through software. Users request and make commitments, collaborate on and complete projects, and develop new software systems through software-based conversations. Yet, software designers and developers approach designing conversation software as a series of generic submissions, rather than as an iterative and reflexive process of specific and varied types of speech-acts. This paper examines two pieces of software: The Coordinator and the Uber Partner (driver) app, and a summary of the dialog surrounding the release of the Coordinator as an implementation of Language/Action Theory.

Min Kim, Sarah-Marie Foley                                         
IxDA 2017    
Findings from the globally upheld and attended Interaction Design conference at New York City, discussing the role of design in healthcare, social innovation, politics, and inclusivity.

Min Kim, Tracy Potter                                        
Design: User Experience from Lisbon    
Findings from a UX conference in Lisbon, Portugal, that covers Conversation UI, ethical designs, and the values of MVPs.

Drama

K. Jenna Ferree                                        
Lighting Design International    
Experience and compare the design technology industries most advanced software and hardware for use in cutting edge Lighting Design.

Daniel Hirsch                                        
ID, Please: An Opera about Border Security    
ID, Please is a new 40-minute chamber opera about immigration and border security that represents a collaboration between the School of Music, School of Drama and Heinz College School of Arts Management. Created by two artists whose work explores issues of identity, with the libretto by queer Jewish-American playwright Daniel Hirsch (MFA in dramatic writing) and music by British-Iranian composer Soosan Lolavar (a visiting Fulbright Scholar at the School of Music), this opera was written in direct response to anti-immigrant rhetoric permeating the UK and USA in recent years. It was produced in collaboration with Carnegie Mellon University, helmed by Heinz arts management student Kevin O'Hora, and premiered at Pittsburgh Opera on April 1, 2017 and has been accepted to perform again at Tete a Tete, an internationally recognized festival of contemporary opera.

Sara Lyons                                        
I'm Very Into You    
I'm Very Into You is a performance adaptation of the published email correspondence by Kathy Acker and McKenzie Wark. The piece is two weeks of emails between Sydney, Australia and San Francisco immediately following an intense 4-day love affair. It is an early artifact of digital relationships, and a brilliant and painfully personal investigation into sex, gender, and power by two brilliant queer writers. Acker, a legendary postmodern punk feminist novelist, passed away in 1997, and this piece is also a rare glimpse into her life, and an opportunity to proliferate appreciation of her transgressive work.

Jessica Medenbach                                        
Les Damnes    
For my research, I accompanied Tal Yarden to Avignon, France and Salzburg, Austria to assist on a theatrical production at the Avignon Theater Festival and a new opera at the Salzburger Festspiele. Through these experiences, I learned After Effects and Isadora and the workflow for producing media for these kinds of productions.

Dan Sakamoto                                        
Electromagika    
Electromagika is an experimental remix of occult practices using modern consumer technology as both interface and character.

Kelly Simons                                        
Early Technical Direction    
Early Technical Direction is a brief overview of my first year of learning to become a theatrical technical director. I participated in a panel at USITT in which early career TDs gathered to share experiences and networking options  

Evan Smith                                        
United States Institute for Theatre Technology    
I will be attending the USITT conference in March of 2017.  This conference provides many classes and networking opportunities to young theatre professionals, as well as demonstrations of upcoming technologies that will be available in the industry.

Music

Matthew Belliston                                        
Choral Conducting: Outside of the Gesture    
Focusing on the research by Christopher Kiver and Peggy Dettwiler in choral conducting outside of the actual gesture. It includes selecting repertoire, posture, preparation, and inviting the singers to become better musicians.

Chen Liang                                        
A Comparative Evaluation on Statistical Models for Score Following    
The project to be funded is a requirement for the Master of Science in Music and Technology degree program. It is a component of my master’s thesis, which is a comparative evaluation of statistical models for score following. Score following is a task that synchronizes a piece of symbolic music notation (such as the music score) with its corresponding audio performances in realtime. The two statistical models to be evaluated are the Hidden Markov Model (HMM) and a Bayesian­based model (called Grubb’s model in the proposal). The motivation to evaluate these systems is that their performance has never been compared quantitatively. My project includes building two systems based on these well­-performing models, and developing a procedure for evaluating score­following systems.

Chen Liang                                        
17th ISMIR Conference Funding    
The 17th ISMIR conference is the world’s leading research forum on music information retrieval (MIR), including processing, analyzing, searching music­related data. My graduate program focuses on music­related technologies, and the MIR is an important branch in music technology. My master thesis’s topic is score following, which is a main topic in the MIR field. The ISMIR conference will invite professionals and researchers covering all topics in MIR for giving poster sessions and tutorials. Participating the ISMIR conference will largely increase career possibilities in music technology industry.


English

Mary Glavan                                        
Parent Advocacy in Special Education Practice: Solving problems in the Face of Personal and Professional Differences    
Using interviews with parent advocates of children with disabilities, this study explores how parents enact rhetorical agency, even as they struggle to meet institutional demands for expertise, knowledge, and authority. I demonstrate how advocates strategically use language to mediate non-normative conditions of public participation, such as parent advocates’ deep personal commitment to and emotional investment in their children.

C.P. Moreau                                        
From College to the Cubicle: A Multiple-voiced Inquiry into the Literate Practices of Recent College Graduates Entering the Professional Work-place    
New research offers three ways undergraduates leaving CMU can successfully transition into the professional workplace as part of a literate practice.    

Doug Phillips                                        
Approaches to Teaching Grammar    
A common complaint among educators and employers is that college students or recent graduates "don't write well." This (inaccurate) perspective is often attributed to STEM students, in particular. As a result, many educators in non-STEM disciplines have begun offering grammar instruction. This report summarizes new approaches to teaching grammar to college students, including online modules for law students and math-based perspectives. These discipline-specific approaches may be promising, but are they effective for teachers who work with students from a range of disciplines, like teachers of technical writing?

Bret Vukoder                                        
Selling Truth Through Violence: The War Documentaries of the United States Information Agency    
My project situates and analyzes the war documentaries employed by the United States' Cold War-era propaganda agency, the USIA. By focusing on films that represent violence, warfare, and atrocity, I seek to explore the lengths to which the agency sought to confirm its projected ethos of "truth," a contradictory, if not paradoxical rhetorical strategy within the confines of propagandistic media.

History

David Busch                                        
The University of the Movement    
This chapter looks at the educational practices and role of students in the civil rights movements and their influence on student thinking about academic education and the role of the university in society.

Susan Grunewald                                        
Assessing the Economic Contribution of German Prisoners of War in the Soviet Union, 1941-1956    
This was a paper presented at the International Business and Economics Conference (IBEC) at the Austral University of Chile at Puerto Montt, Puerto Montt, Chile. Held for 11 years after the end of the war and 7 years longer than the other Allied victor nations, the Soviets kept their German POWs through 1956 to facilitate postwar reconstruction. Only once the economy recovered were the remaining POWs held as bargaining chips for Cold War diplomacy.

Zhaokun Liu                                        
Negotiation on the Repatriation of U.S. Soldiers' Remains From North Korea Under the UN Banner    
The recovery of the remains of the American servicemen killed in the Korean War is not a task just for the U.S.  It is actually the UN Command that was in charge of negotiating with North Korea for recovering the bodies. This project investigates why the UN only received a small number of remains in 1954 and then none until 1990. I contend that early negotiations between North Korea and the UN exacerbated their mutual distrust, making a cooperation in recovering the remains impossible. North Korea commandeered the negotiations for its political propaganda that the UN could only passively defend.

Matt Nielsen                                        
Amphibious Flight and Transboundary Water Politics: Runaway Slaves in the Lower Orinoco River Basin in the 18th Century    
Using colonial records and missionary documents, this paper analyzes the experiences of runaway slaves who fled the Dutch colony of Essequibo and entered the province of Spanish Guayana in the 1700s. I argue that these runaways pursued self-emancipation in discrete ways and contributed to and benefitted from a network of lasting resistance and a collective body of knowledge concerning amphibious flight in the Orinoco river basin. Furthermore, by forging alliances with local Indians, sharing navigational skills, petitioning colonial Spanish officers to defend their freedom, pursuing economic activities, etc., the runaways helped shape the region’s transboundary politics.  

Meredith Soeder                                        
The Undetermined Place of Jazz in Post-World War II German Art Music    
The resurgence jazz music’s popularity among enthusiastic youth and some open-minded Germans brought added pressure to post-World War II Germany’s dormant art music world. The role of jazz in art music became a hot topic in debates about the future of German music. These debates encompassed an interwoven discussion of the value of jazz, whether or not jazz could be taken seriously, its role highbrow versus lowbrow music, and jazz’s racial identity. For these critics, composers, and educators, the relationship between art music and jazz was still unresolved, but jazz did indeed regain lost ground in compositions and concert halls.

Modern Languages

Tianxu Chen                                        
Lexical Inference in L2 Chinese: An Investigation of Morphological Awareness, Word Semantic Transparency and Contextual Cues    
The present study aimed to investigate how L2 learner-related variables (i.e., morphological awareness) play a role in lexical inference interactively affected by L2 language-related (i.e., word semantic transparency) and task-related (i.e., contextual cues) factors. Nighty collegiate Chinese learners whose instructional level was intermediate participated the study. They completed five tasks in Chinese. The findings showed that (1) word semantic transparency and contextual cues independently and interactively affect L2 Chinese lexical inference; (2) compared to low-achieving learners of morpholgocial awareness, learners with high morphological awareness performed better on lexical inference under different conditions.

Qiong Li                                        
Use of Mitigation Devices in Heritage Learners of Chinese    
This study investigated pragmatic competence of Chinese heritage learners (CHLs) on the use of three Chinese-specific mitigation devices: Chinese sentence final particles, yixia/xia “for a while”, and reduplication of verbs, in comparison with foreign language learners of Chinese (CFLs) and native Chinese speakers (NSs). Participants included 60 Chinese language learners in the intermediate and advanced-level classes (31 CHLs and 29 CFLs). They completed a computerized written production task that involved two types of situations. The statistical findings showed no notable significant heritage advantage; however, the post-hoc analysis revealed the unique nature of CHLs’ pragmatic competence, vacillating between CFLs and NSs.

Maria Pia Gomez Laich                                        
Comprehension of indirect Meaning in Spanish as a Foreign Language    
This study investigated comprehension of indirect meaning among learners of L2 Spanish via an original computer-delivered multimedia listening test. The listening test assessed comprehension of three types of indirect meaning: indirect refusals, indirect opinions, and irony. Participants’ comprehension was analyzed for accuracy and comprehension speed.   

Tianyu Qin                                        
Computerized Dynamic Assessment of L2 Chinese Pragmatic Comprehension    
This paper reports the results of a study that extended computerized dynamic assessment (C-DA) to the domain of second language (L2) pragmatics. Drawing on Vygotskian psychology, C-DA integrates automated instruction (i.e., preprogrammed prompts) as part of a standardized web-based testing procedure. The focus is on comprehension of conversational implicature in L2 Chinese (i.e., implied or indirect meaning), using materials from the interlanguage pragmatics literature (Taguchi, Li, & Liu, 2013) that we adapted for the C-DA format. 67 US university learners of Chinese participated in this study.  

Xiaofei Tang                                        
Use of Technology for Learning L2 Pragmatics    
Digitally-mediated technology provides great potential for second language (L2) pragmatics learning, as it creates authentic contexts for learners to practice their L2s. Despite researchers’ growing interest in technology-enhanced learning, the exact efficacy of using technology in L2 pragmatics learning still remains unknown. This synthesis examines the role of technology in L2 pragmatics learning and synthesizes the learning outcomes in previous empirical studies. Two research questions were addressed: 1) How is technology used to develop L2 learners’ pragmatic competence?; and 2) What are the pragmatics learning outcomes and how are they documented? A total of 21 studies were selected after an exhaustive literature search and applying certain inclusion/exclusion criteria. The findings revealed three ways in which technology had been used: as an instructional tool, as a medium for naturalistic interaction, and as both. The findings also showed positive results of using technology to facilitate pragmatics learning. This synthesis presented how the pragmatics learning outcomes were measured and analyzed differently in previous studies. This paper further discusses how different ways of using technology reflect different conceptualizations of pragmatic competence, and generates implications for future research on technology-enhanced pragmatics learning.  

Aurora Tsai                                        
Fostering and Monitoring Higher Order Reading and Learning Skills in the Foreign Language Classroom    
This study investigated how we can assess the integrating prior knowledge with text information and whether or not it promotes learning cross-cultural learning among Japanese foreign language learners in university classes. 

 

Philosophy

Remco Heesen                                        
When Journal Editors Play Favorites    
Should editors of scientific journals practice triple-anonymous reviewing? I consider two arguments in favor. The first says that insofar as editors' decisions are affected by information they would not have had under triple-anonymous review, an injustice is committed against certain authors. I show that even well-meaning editors would commit this wrong and I endorse this argument.   The second argument says that insofar as editors' decisions are affected by information they would not have had under triple-anonymous review, it will negatively affect the quality of published papers. I distinguish between two kinds of biases that an editor might have. I show that one of them has a positive effect on quality and the other a negative one, and that the combined effect could be either positive or negative. Thus I do not endorse the second argument in general. However, I do endorse this argument for certain fields, for which I argue that the positive effect does not apply.

Aidan Kestigian                                        
Using Blogs in the Political Theory Classroom    
In the past decade, several professors have advocated for the use of blogs in undergraduate courses in philosophy, arguing that blogs are beneficial for student learning, as blogs are forums for student collaboration and engagement with course material outside the classroom. In this paper I argue that blogging assignments can be beneficial for introductory-level undergraduate courses in philosophy for two reasons yet to be fully explored in the pedagogical literature. First, blogging assignments can act as low-stakes practice for paper writing. Second, blogging assignments give students the freedom to explore the relevance of course content to real world problems.

Psychology

Brian Chin, Seh-Joo Kwon, Sandy Chen, Morgan Morrison, Gowri Sunder, Sheldon Cohen, David Creswell    
A Novel Non-verbal Measure of Emotions    
One hundred and twenty (N = 120) undergraduates were randomly assigned to view one of four 10-minute blocks of film clips: high arousal, positive valence (humor); high arousal, negative valence (horror); low arousal, positive valence (content); low arousal, negative valence (sad). Participants were asked to rate their current mood both before and after film viewing in two ways: (i) by squeezing a handheld dynamometer that measures grip strength; and (ii) by using a 9-point paper Likert scale. We predict a moderate-strong correlation between dynamometer and paper ratings of mood, as well as significant between-group differences in both types of ratings.

Cassandra Eng                                        
Infant Attention & Corresponding EEG: Potential Indicators of Childhood AD/HD    
Infant attention and corresponding brain-behavior associations are understudied in the development of AD/HD. Abnormal brain activity is a candidate factor associated with brain disorders, and the electroencephalogram(EEG) is a non-invasive method for examining abnormalities in development. Retrospectively, attention and brain activity at 5 months of age of 19 infants diagnosed at childhood for AD/HD were compared to those of 19 matched control infants. There were findings of decreased brain activity in AD/HD infants, emphasizing a deficit in neural activity. Infant attention and EEG as early as 5 months, can potentially be used as indicators of childhood AD/HD.

Katilyn Mascatelli                                        
Impact of Race, Class, and Gender on Person Perception    
Research shows that the attributes associated with the category “women” overlap more with attributes associated with “white women” than “black women.” Similarly, attributes associated with “men” overlap more with attributes associated with “white men” than “black men.” Thus, when people think about gender they are really thinking about gender in white people. Similarly, stereotypes of white people (e.g., intelligent, ambitious, well-spoken) overlap more with the stereotype of upper- than lower-class people. Stereotypes of black people (e.g., lazy, unreliable, ignorant) overlap more with the stereotype of lower- than upper-class people. Thus, when people think about white people they think about upper-class people. A series of studies explores the possibility that gender stereotypes only apply to white, upper-middle class people.  

Vencislav Popov                                        
Semantic-episodic Interactions During Memory Retrieval    
Semantic and episodic memory retrieval are supported by two largely overlapping neural networks. However, most studies have compared them separately to non-mnemonic tasks, which prevents us from inferring how parts of these networks contribute differentially to the retrieval of semantic vs episodic information. In an fMRI experiment participants made either semantic or episodic judgments on the same stimuli. In the first phase, participants respondedwhether facts about famous people were true or false. Subsequently, participants responded whetherintact or recombined facts from the first phase weretested or not in the first phase. The behavioral results suggested that semantic and episodic information interact during bothretrieval conditions. Univariate analyses revealed greater activation of theangular and middle temporal gyri in the semantic task, and greaterparahippocampal and anterior hippocampal activation in the episodic task. Results are discussed with respect to the interaction between semantic and episodic information during memory retrieval.

Casey Roark                                        
The Interaction of Perceptual Dimensions in Auditory Category Learning    
Research on category learning distinguishes category distributions that require selective attention to one dimension and those requiring integration across multiple dimensions. Our goal was to examine the influence of the dimensions defining categories on learning these categories. Across training accuracy, generalization to novel exemplars, and decision bound computational models that approximate strategy use, there was a bias to integrate across dimensions rather than to selectively attend to either dimension. This bias was eliminated when the category boundary was rotated to reflect a negative relationship between dimensions. These findings demonstrate the importance of understanding perceptual, as opposed to physical, dimensions and their interaction during category learning.

Melissa Zajdel                                        
Illness Identity and Centrality    
Illness identity is the extent to which an individual does or does not incorporate his or her illness into their sense of self. This study sought to determine the potential relations between illness identity and centrality to adjustment outcomes, as well as identify potential moderators that may affect these relations.

Social and Decision Sciences

Nikolos Gurney                                        
Filling in the Blanks: What Customers Assume About Potentially Valuable, but Missing, Information    
Information asymmetry is a central concept in the study of information economics.  Sellers often have more and better information about their goods than buyers which creates an environment where sellers can exploit unwitting buyers.  Although game theoretic analyses suggest that a simple unraveling logic can save a buyer from falling victim to unscrupulous sellers, converging evidence from econometric analyses and experimental economic games shows that people do not unravel market choices.  In a series of hypothetical choice experiments, we first establish that the normative prediction of unraveling does not occur in a dining out scenario and then test three different interventions to increase normative behavior from participants in similar scenarios.  We find that providing choice context, e.g. other options that vary on important dimensions, helps to improve choice behavior.

David Hagmann                                        
The Influence of Loss Aversion on Explore-Exploit Decisions    
In “explore-exploit” situations, decision makers must choose between exploring unknown options and exploiting known options.  We study how explore-exploit decisions vary under the influence of loss aversion, predicting that (1) people will be less likely to explore if doing so can lead to losses and that (2) people will be less likely to exploit when doing so would lead to repeated losses.  To examine these predictions, we used a novel multiple round computer task in which participants explored one of two one-dimensional environments that were equivalent in terms of potential payoffs.  In one decision environment, participants encountered only gains in all rounds of play.  In the other, they were given an up-front payment that was offset in subsequent rounds, where participants experienced gains or losses.  Across multiple studies, we find evidence for both of our predictions.  Additionally, we demonstrate that loss aversion can be adaptive, leading participants in low payoff environments to higher total rewards.

Ania Jaroszewicz                                        
Yes I Can: Agency Moderates the Effects of Scarcity and Other Adverse States on Decision-Making    
Research shows adverse states such as scarcity of money or time can negatively impact decision-making. Critically, individuals in such states often lack agency over transitioning to a less adverse state. We propose that perceived agency moderates the effect of adverse states on behavior involving intertemporal tradeoffs, e.g. savings and education. We first demonstrate that exposure to resource scarcity increases impatience, but that endowing people with agency over the scarcity moderates its impact on decision-making, such that those with agency appear just as patient as those not experiencing scarcity at all. This effect occurs even when the agency is not exercised.

Talya Lazerus                                        
Beyond Information: The Role Of Emotion And Hunger In Perceptions Of Caloric Content    
Most food behavior interventions are designed to increase awareness of nutrition information. However, these interventions are developed with assumptions about the link between perceptions of nutritional content and the processes behind people’s food choices. In this paper, we explore what factors affect these processes and what information individuals gather without explicit nutritional information being provided. Specifically, we investigate how calorie estimations vary based on people’s internal states (i.e., hunger level, emotions). Analyzing how people perceive and understand such information may lead to better crafted interventions and policies.

Statistics

Jisu Kim                                     
R Package TDA for Topological Data Analysis    
This poster gives an introduction to the R package TDA, which provides some tools for Topological Data Analysis. The salient topological features of data can be quantified with persistent homology. R package TDA provide an R interface for the efficient algorithms of the C++ libraries GUDHI, Dionysus, and PHAT, including functions for computing the persistent homology. The R package TDA also includes an algorithm for the cluster tree that corresponds to the density clustering.

Alan Mishler                                        
Filtering Tweets for Social Unrest    
Since the events of the Arab Spring, there has been increased interest in using social media to anticipate social unrest. While efforts have been made toward automated unrest prediction, we focus on filtering the vast volume of tweets to identify tweets relevant to unrest, which can be provided to downstream users for further analysis. We train a supervised classifier that is able to label Arabic language tweets as relevant to unrest with high reliability. We examine the relationship between training data size and performance and investigate ways to optimize the model building process while minimizing cost. We also explore how confidence thresholds can be set to achieve desired levels of performance.

Alex Reinhart                                        
Point Process Modeling with Spatiotemporal Covariates for Predicting Crime    
Extensive research has shown that crime tends to be concentrated in hotspots: small pockets with above-average rates of crime. Criminologists and law enforcement agencies want to better predict crime hotspots and understand the factors that cause them, in order to target interventions. Prior research suggests that past crime hotspots, spatial features (like bus stops or bars), and leading indicators (like 911 calls) are all predictive of future crime, but no proposed predictive policing model can account for all of these factors. We have adapted a previous self-exciting point process model to incorporate past crime data, leading indicators, spatial features and spatial covariates (like population density or zoning data), and developed new tools to evaluate the performance of the model and select variables. We demonstrate the model on five years of Pittsburgh crime data.

Giuseppe Vinci                                        
Auxiliary Regularization of Cortical Covariance    
One of the most important challenges of computational neuroscience is estimating functional connectivity, that is inferring dependence structure among neural signals. Nowadays neuroscientists can record the activity of hundreds of neurons simultaneously, but only on limited numbers of trials. Sparse Gaussian Graphical Models (GGM), such as the Graphical Lasso, can provide sparse dependence structure estimates, but their performance in neural data can be unsatisfactory. We propose regularized GGM that incorporate neurophysiological information and provide better connectivity estimates. We apply the methods to spike count data recorded with multielectrode arrays implanted in macaque visual cortex areas V1 and V4.

Entertainment Industry Management

Laila Archuleta                                        
Producers Guild of America Conference    
Attending speaker panels and meeting other industry heads.

Information Systems & Management: Information Security Policy Management

Jason Bartolacci                
Social Engineering in Action: Beating Contemporary Information Security Measures with an Old-fashioned Phone Call    
Social engineering poses a significant threat to financial institutions, organizations and corporations worldwide.  This presentation will provide an overview of social engineering and will discuss the highlights of several cyber financial fraud cases that involve social engineering.  Importantly, this presentation will also discuss methods to combat social engineering.

Information Systems & Management: Information Systems

Arslan Aziz                                        
Evaluation of Marketing Segments' Significance and Their Attributes by Identifying the Most Important Features and Performing Segment Valuation Using Those Features.    
We analyze the impact of third-party tracking on competition in targeted advertising.

Inchara Bellavara Diwakar                                        
On Extracting Features from Asynchronous Multivariate Data Streams    
We study the problem of extracting predictive features from asynchronous multivariate data streams.We propose a smoothing technique for segmenting measurements collected at irregularly spaced time points into trend and value abstractions. This process allows us to transform a measurement sequence into variables that be supplied to prediction models. We show that using these abstracted temporal features in building prediction models improves predictive accuracy over models that solely consider summary statistics such averages and counts.

Nirjhar Bera                                        
Enabling Housing Connector Program to Serve Individuals with Disabilities Efficiently    
The Housing Connector program is a collaboration between the ACTION-Housing, Inc. and the Allegheny County Department of Human Services. Its goal is to map housing options for individuals with disabilities. The problem is that the currently implemented technology will not be able to support the increasing rate of client enrollment. Activities, like tracking client information and generating performance report, is becoming laborious and inefficient. My objective is to improve their efficiency by implementing an innovative and sustainable technological solution. This solution will directly impact and enhance the housing search process for disabled homeless people.

Lujie Karen Chen                                        
Riding an Emotional Roller-coaster: A Multimodal Study    
Solving challenging math problems often invites a child to ride an ``emotional roller-coaster" and experience a complex mixture of emotions including confusion, frustration, joy, and surprise. Early exposure to this type of ``hard fun" may stimulate child's interest and curiosity of mathematics and nurture life long skills such as resilience and perseverance. However, without optimal support, it may also turn off child prematurely due to unresolved frustration. An ideal teacher is able to pick up child's subtle emotional signals in real time and respond optimally to offer cognitive and emotional support. In order to design an intelligent tutor specifically designed for this purpose, it is necessary to understand at fine-grained level the child's emotion experience and its interplay with the inter-personal communication dynamics between child and his/her teacher. In this study, we made such an attempt by analyzing a series of video recordings of problem solving sessions by a young student and his mom, the ideal teacher. We demonstrate a multimodal analysis framework to characterize several aspects of the child-mom interaction patterns within the emotional context at a granular level. We then build machine learning models to predict teacher's response using extracted multimodal features. In addition, we validate the performance of automatic detector of affect, intent-to-connect behavior, and voice activity, using annotated data, which provides evidence of the potential utility of the presented tools in scaling up analysis of this type to large number of subjects and in implementing tools to guide teachers towards optimal interactions in real time.  

Jamie Diner                                        
Audience Discovery: Valuation of Segments and Attributes    
Evaluation of marketing segments' significance and their attributes by identifying the most important features and performing segment valuation using those features.

Abhinav Maurya                                        
Making Personalized Skill Recommendations using Bayesian Member-Job Matching    
The central premise of our research is that increasing the productivity of a member of the workforce (and thereby of the economy as a whole) crucially depends on identifying skills whose acquisition will yield the highest utility gains for that member. To this end, we develop a novel attribute-based Bayesian matching model BayesMatch to match members to other similar members as well as relevant jobs, which is followed by a skill recommendation step SkillR that makes utility-based skill recommendations to members. Our extensive quantitative evaluation using a rich dataset comprised of professional profiles and job postings from LinkedIn suggests that skill recommendations made by our algorithm SkillR have higher correlation with skills demanded in heldout future jobs than those made by a traditional collaborative filtering algorithm SkillRBaseline that does not utilize information about skills demanded in jobs.

Christopher Worley, Justin Cole, Ben Simmons                                        
Students for Urban Data Systems - Code for America Summit    
Students for Urban Data Systems (SUDS) is a student group that brings together the vast array of disciplines that contribute to the conversation of how cities are adapting new methods of “big data” collection and analysis to improve their service delivery, from providing water and police protection to repairing infrastructure. SUDS group members attended the Code for America Summit, which focused on civic technology.

Information Systems & Management: Information Technology

Chad Hicks                                        
NYC 3D Printing Expo Report    
A review of 3D printing from an IT / Information Security mindset.

Public Policy & Management: Arts Management

Patrick Zakem                                        
The 2016 Humana Festival of New American Plays    
First held in 1977, The Humana Festival of New American Plays is the nation's preeminent theatre festival celebrating contemporary voices in the American theatre. Each year, theatre lovers, critics, producers, and literary managers from around the world come to Louisville, Kentucky to attend world premiere productions of new plays and see the new work being generated by America’s most promising artists. Each year the festival culminates in a Professional’s weekend, which combines the plays with panels, networking events, and discussions.  I attended the 2016 Festival during one of these weekends, and am presenting my experience.

Public Policy & Management: Creative Enterprises

Jane Bowers, Zining Xie                                        
An Updated Picture of U.S. Dance Internationally    
At the Association of Performing Arts Presenters conference in 2009, American Dance Abroad (ADA) presented the results of 18 months of research on the prevalence of U.S. dance touring internationally.  By using the 2009 survey with modifications reflecting contemporary concerns, the Systems Synthesis team determined the current needs of touring dance companies and artists in order to recommend strategies for ADA to best serve the community.

Mengdi Ding, Zaijun Wu                                        
Emerging Technologies for Museums    
Digitally mediated personalization and personalized learning are two global prominent trends in museums in recent years. Many leading museums in the world have embarked on using emerging technologies to enhance visit experience. Our research provides a view into AR (Augmented Reality) and AI (Artificial Intelligence) technologies and opportunities for their uses in museums.

Annesha Ganguly                                        
How APAP Helped Me Build a Network    
I will be talking about how a conference is helpful in networking and getting to know the way the industry works.

Beth Geatches                                        
Association of Performing Arts Presenters 2017    
Thousands of presenting organizations, artists, managers, and arts leaders from across the globe attend the Association of Performing Arts Presenters Conference in New York City. Not only is this an excellent opportunity for networking and finding summer internship opportunities within the field of Performing Arts Presenting, the professional development sessions available and volunteering work is an unmatched hands-on learning opportunity for early career arts management professionals.   

Justin Gilmore                                        
Nonprofit Technology Education Network Conference    
The Nonprofit Technology Conference (NTC) is the nonprofit sector’s signature technology event. We assemble over 2,000 of the best and brightest nonprofit professionals from around the world. Together, they collaborate, innovate, and discover new ways to spark change with technology.

Amelia Nichols                                        
Austin Film Festival Development Plan    
Development plan proposal for Austin Film Festival

Anne Marie Padelford, Jess Bergson, Mandy Ding, Kate Martin, Brett Crawford
Fueling Change via Smart Engagement with Existing & New Technologies    
With the rise of new digital tools such as Virtual Reality and Augmented Reality, arts marketers are under pressure to  keep their organizations “technologically-relevant.” However, many organizations are instead finding themselves increasingly irrelevant in the digital age. This workshop explored the value of the live artistic experience that co-exists alongside the digital age. Participants gained a foundational understanding of how emerging IoT and AR/VR technologies work and the tools for determining a technology strategy best suited to meet a creative community.  This workshop/panel was presented at the 2016 National Arts Marketing Projected Conference in Austin, TX.

Jessica Tkach, Marshall Bain, Jessica Yang, Yasmin Foqahaa, Ellen Murphy, Dervla McDonnell        
Millennial Engagement at The Phillips Collection    
The Phillips Collection, a time-honored modern art museum, parallels the industry wide concern for engaging younger generations. Confronted with issues of audience diversity, participation and retention, the Phillips seeks to understand how to best steward young people and cultivate the next generation of lifelong Phillips patrons. Our team will evaluate both the real and perceived barriers of the Phillips in order to broaden public engagement and help the museum meet the demands of younger stakeholders.     Research will begin with auditing and benchmarking the current position of the Phillips against DC specific alternatives. This step will help orient the research team around real and perceived barriers, current and future trends in demand, and overall brand perception. Through the administration of surveys, audience profiles, and design thinking research methodologies, the Phillips can begin to frame a new cycle of constituents while also learning how to best reach and serve their current audience.   

Jonathon Weber                                        
Of Value    
The for-profit visual arts industry is nuanced. Like the luxury goods market artwork and antiques are traded between parties in markets like any other commodity. However, unlike commodities, art and antiques have inherent emotional qualities which are hard to justify to any end rationally. Unlike, luxury goods, art & antiques are also inherently unique, with very limited quantities, therefore there is little precedent to govern systems. From an environment that is unregulated, highly lucrative, and challenging to measure in an intelligent fashion the agency role is fulfilled through Appraisers. Its industry, community and tools will be the subject of survey.

Zining Xie                                        
American Dance Abroad Synthesis Project    
Based on a survey conducted by American Dance Abroad in 2009, our team designed a modified survey to collect data on the international touring practices and needs of U.S. dance companies and artists over the past three years. The survey consists of three main sections: 1) Company Profile, 2) Touring Practices, and 3) Touring Needs. Additionally, the survey tracks specific changes between 2009 and 2016. The survey was distributed to professional dance companies and independent dance artists across the United States.

Shana Xu                                        
Producers Guild of America Conference: Produced By New York    
Key takeaways and learnings from "The Truth Sells: Distributing Your Documentary and Building Your Audience" "The Storytelling Horizon: Digital Possibilities for Producers" "Meet the New Digital Buyers" and "Demystifying the Money: Straight Talk about Film Financing" panels.


Public Policy & Management:Public Policy & Management

Mariah Farbo, Justin Cole, Krista Kinnard, Lauren Renaud                            
Do Good Data    
Several Heinz students from the data analytics track went to a conference on applying data to social good. We will present our findings from the conference and how we applied what we learned in our education.

Brian Gillikin                                        
The GW/World Bank October Conference on Global SME Growth & Innovation    
SME (Small & Medium Business) growth is an increasingly important indicator of economic growth worldwide, especially in conjuncture with localized innovation. At this conference, experts presented on a wide range of global contexts where SME dynamics create and/or hinder innovation and growth.

Britta Glennon                                        
Does Offshoring Manufacturing Harm Home Country Innovation? Evidence from Taiwan    
The consequences of relocating manufacturing functions abroad for firms’ ability to innovate are not well understood in the literature. On the one hand, some of the literature contends that offshoring can have a positive effect on home country innovation through efficiency gains and reverse technology transfer. However, other strands of the literature argue that separating the manufacturing and R&D functions of a firm can be dangerous to a firm’s innovative capacity, and hence also to the firms’ home country. We provide some empirical evidence clarifying this debate through this project. In particular, we study the impact of Taiwanese high-tech companies’ decision to offshore manufacturing to mainland China on their patenting behavior. We exploit a policy shock in Taiwan in 2001 that lifted many of the restrictions that had prohibited Taiwanese companies from legally offshoring their manufacturing to China prior to that date using a 2SLS strategy to get at the causal relationship between offshoring and innovation. The response of Taiwan’s electronics and IT firms to this policy shock was rapid and substantial – a large fraction of these firms’ manufacturing operations shifted to mainland China within just a few years.  How did this affect the level of firm innovation? Using a unique and highly granular panel dataset, we find that offshoring has a negative impact on a firm’s innovation.

Guangwei Li                                        
Bridge to Excellence? The Impact of International Coinvention on Multinational R&D in China    
Multinational corporations are significantly increasing the amount of R&D they conduct in emerging economies. Many of the patents generated by this R&D are the result of international coinvention—that is, they are produced by an international team of inventors, with some inventors residing in the emerging market in which an R&D subsidiary is based and other inventors located elsewhere, usually in an advanced industrial country. These patterns are particularly pronounced in China. We hypothesize that multinationals use international coinvention as a way of connecting the high level of basic engineering talent their Chinese recruits possess to the sophisticated understanding of the global technological state-of-the-art developed by their staff in advanced industrial economies. We also hypothesize that repeated participation in international coinvention builds the skills of local employees to the point where they can produce high-quality inventions without the direct intellectual input of R&D staff located outside China. We test these hypotheses and find strong support for both.   

Yue Qiu                                        
Sustainatopia 2016    
Overall, this conference was a good learning opportunity for me, as the conference brings together actors across social, financial and environmental sustainability sectors to comprehensively discuss solutions and responsibility towards sustainable development. The conference also offered seminars on specific areas of my interest, such as Sustainable Strategies and Smart Cities.

Filipa Reis                                        
The Impact of Time-shift Television on TV Viewership Behavior    
We study the impact of Time-Shift TV (TSTV) on TV consumption using data from a randomized experiment in which a telecommunications provider gifted a set of new entertainment channels to a sample of its subscribers. A random subset of the sample received the channels with the TSTV feature and another random subset received the channels without TSTV. Using difference-in-differences we find that, TSTV lead to an increase of total TV time without reducing live TV time while also leading households to shift their consumption towards more popular programs. Finally, we show that TSTV was not used to strategically avoid ads.




Biological Sciences

Surya Aggarwal                                        
Regulation & Characterization of a Novel Pneumococcal Peptide Implicated in Biofilm Development    
Streptococcus pneumoniae (pneumococcus) is a causative agent of invasive and non-invasive diseases. During infection, pneumococcus organizes itself into complex biofilm communities. While previous work has shown a correlation between upregulation of the competence pathway and biofilm development, these questions still remain largely unanswered. We have identified a putative secreted peptide, BAP (Biofilm Associated Peptide) that is upregulated by a competence-associated regulator and induced in chinchilla middle-ear infection relative to planktonic cultures.. In vitro, deletion mutants of bap display decreased biofilm biomass and thickness, and increased biofilm roughness in mature biofilm structures.  These findings are consistent with a role for BAP in the regulation of biofilm development.

Rolando Cuevas                                        
A Novel Streptococcus Pneumoniae Cell-cell Communication Peptide is a Virulent Determinant Involved in Early Biofilm Development    
Streptococcus pneumoniae asymptomatically colonizes the human upper respiratory tract and disseminate to other tissues, most commonly the lungs and middle ear, causing mild to severe disease. The mechanisms by which S. pneumoniae senses and responds to environmental signals within the host are poorly understood. Here, we describe for the first time the previously uncharacterized small peptide VP1 and its role in biofilm development and pathogenesis in a clinically relevant strain. Our findings will shed light on the biological processes that lead to microbial virulence.  

Katherine Lagree
C. Albicans Biofilm Formation
Candida albicans is an opportunistic pathogen that can form drug resistant biofilms on implanted medical devices. Numerous transcriptional regulators have been shown to be required for biofilm formation. Here we describe a transcription factor, Zfu2, that controls biofilm formation through a distinct pathway from other well-described biofilm regulators.  We found that Zfu2 controls the expression of several related CFEM genes (CSA2, PGA7, RBT5) that function in heme iron acquisition. Our results suggest that the biofilm defect of a zfu2 mutant is due to its inability to acquire iron from heme from the host. Our results also suggest that heme iron may be the sole source of iron in the in vivo rat venous catheter model of biofilm formation.  To understand how Zfu2 fits into the greater iron regulatory network of transcription factors, we have begun to profile transcription factor mutants under iron limiting conditions or with heme as the sole iron source.         


Chemistry

Sikandar Abbas                                        
Effect of Plasmonic Substrates on Photo-stability of Organic Layers    
Realization of highly efficient and cost effective organic photo-luminescence applications require deep understanding of photochemical interactions. Photo-damage of emissive organic layers is one factor that decreases their overall efficiency and longevity. Here, the effects of metal film substrates on  the emission properties of organic layers is investigated using total internal reflection (TIRF) fluorescence microscopy. It is found that the OPPV-13 layer exhibits a remarkable increase in photo-stability when deposited on thin gold films relative to that on glass, even in the presence of molecular oxygen (O2) and under laser illumination. Moreover, photo stability behavior is being analyzed when Au films is replaced with non-plasmonic metal substrate.

Christian Legaspi                                        
The Role of Local Environment on the Electronic Properties of a Novel Blue-emitting Donor-acceptor Compound    
We have recently synthesized a novel blue-emitting, donor-acceptor system employing carbazole as the donor and a benzothiazole derivative as the acceptor, BTZ-CBZ. We find that the solution-phase emission of BTZ-CBZ is highly dependent on solvent dielectric, both in lineshape and emission maximum. However, this solvatochromic shift is suppressed when BTZ-CBZ is confined to a rigid matrix. We theorize that solvent reorganization plays a significant role in the relaxation mechanism of the BTZ-CBZ excited state. Using both spectroscopic and computational techniques, we explore this phenomenon to better understand the behavior of BTZ-CBZ as a solid-state blue emitter.

Logan Plath                                        
Characterization of Zinc Oxide Nanoparticles Using Superconducting Tunnel Junction Cryodetection Mass Spectrometry    
Matrix-assisted laser desorption/ionization (MALDI) coupled to a time-of-flight (TOF) mass spectrometer with superconducting tunnel junction (STJ) cryodetection was used to characterize synthetic zinc oxide (ZnO) nanoparticles with varying ligand coatings. Not only can a single mass spectrometry (MS) analysis provide high m/z determination, dispersity, and size, but also ligand loading via STJ energy resolved metastable fragmentation. This methodology demonstrates applicability to many areas of nanoparticle characterization through a single MS experiment.

Tyler Womble                                        
Synthesis, Characterization, and Polymerization of Resonance Stabilized Main Group Polyelectrolytes    
Present global energy production using non-renewable sources (i.e. fossil fuels) have dramatic economic, climate, and health-related impacts. Alternative technologies, such as fuel cells or batteries, harness chemical energy and ion transport phenomena to efficiently produce and store energy that is also environmentally benign. Current research efforts are geared toward improving device performance and market competitiveness in order to replace traditional combustion engines. A key component of these devices is an ion-conducting membrane and our work has focused on developing polymer materials that are thermally and chemically stable, mechanically robust, and have high anion transport.

Qing Ye                                        
Source Contributions and Mixing State of Atmospheric Particulate Matter in Pittsburgh    
We performed mobile sampling in Pittsburgh starting August 2016 to investigate the sources and mixing state of airborne particles. We find that emissions from traffic and restaurant cooking are major contributors of particulate matters in Pittsburgh.             

Physics

Devashish Gopalan                                        
Magnetism and Proximity Effects in Layered Two-dimensional Materials    
Van der Waals materials have a layered structure, and can be made atomically thin by scotch tape exfoliation. The versatility of these materials has allowed the study of diverse, intriguing physics in two dimensions.  Furthermore, by means of proximity effects, one material can acquire the properties of an adjacent material by bringing them in close contact.    I will report on proximity-effect induced magnetism in graphene, using chromium silicon tritelluride (CrSiTe3), which is a layered ferromagnetic semiconductor.

Sukhdeep Singh                                        
Studying the Evolution of the Universe with Weak Gravitational Lensing    
As light rays travel across the universe, their paths are deflected by the gravitational effects of the intervening matter. This phenomenon, Gravitational lensing, distorts and correlates the shapes of galaxies and is an important tool to study dark matter and dark energy over cosmological scales. We use weak gravitational lensing maps constructed with data from large cosmological surveys (SDSS and Planck), to map the dark matter distribution around galaxies at different cosmic epochs. I will present results from our study and discuss their applications to test the theories of gravity and study the evolution of the universe.    

Xiaoou Zhang                                        
Dark Excitons in Gapped Chiral Fermion Systems    
We provide a new mechanism to realize dark lowest exciton state based on gapped chiral fermion model. We found that the angular momentums of bright and dark exciton are related to the winding number of chiral fermion and crystal symmetry. As possible realization of our mechanism, we propose that the lowest exciton states in magnetically doped surface state of topological crystalline insulator (TCI) and gated 3R stacking bilayer MoS2 are optically dark. In the latter case, we further show that a gate voltage can be used to tune the lowest exciton state between dark and bright. This provides a pathway to electrical control of optical transitions in two-dimensional material.


Computer Science

Kristen Gardner                                        
A Better Model for Job Redundancy: Decoupling Server Slowdown and Job Size    
Recent computer systems research has proposed reducing latency via redundant requests. The idea is to replicate a request at multiple servers and wait for the first copy to complete service. Redundancy helps overcome server-side variability---the fact that server speeds are unpredictable and change over time.    For analytical tractability, theoretical work on redundancy often assumes that a job's replicas experience independent runtimes across servers. This is unrealistic and leads to theoretical results that do not match empirical results. We introduce a more realistic model that allows for correlated runtimes, and propose the Redundant-to-Idle-Queue policy, which has provably excellent performance.

Christian Kroer                                        
Dynamic Thresholding and Pruning for Regret Minimization    
Regret minimization is widely used in determining strategies  for imperfect-information games and in online learning. In  large games, computing the regrets associated with a single  iteration can be slow. For this reason, pruning – in which  parts of the decision tree are not traversed in every iteration –  has emerged as an essential method for speeding up iterations  in large games. The ability to prune is a primary reason why  the Counterfactual Regret Minimization (CFR) algorithm using  regret matching has emerged as the most popular iterative  algorithm for imperfect-information games, despite its relatively  poor convergence bound. In this paper, we introduce  dynamic thresholding, in which a threshold is set at every  iteration such that any action in the decision tree with probability  below the threshold is set to zero probability. This enables  pruning for the first time in a wide range of algorithms.  We prove that dynamic thresholding can be applied to Hedge  while increasing its convergence bound by only a constant  factor in terms of number of iterations. Experiments demonstrate  a substantial improvement in performance for Hedge as  well as the excessive gap technique.

Stefan Muller, Ram Raghunathan                                        
Hierarchical Memory Management for Parallel Programs    
An important feature of functional programs is that they are parallel by default. Implementing an efficient parallel functional language, however, is a major challenge, in part because the high rate of allocation and freeing associated with functional programs requires an efficient and scalable memory manager.    In this paper, we present a technique for parallel memory management for strict functional languages with nested parallelism. At the highest level of abstraction, the approach consists of a technique to organize memory as a hierarchy of heaps, and an algorithm for performing automatic memory reclamation by taking advantage of a disentanglement property of parallel functional programs. More specifically, the idea is to assign to each parallel task its own heap in memory and organize the heaps in a hierarchy/tree that mirrors the hierarchy of tasks.    We present a nested-parallel calculus that specifies hierarchical heaps and prove in this calculus a disentanglement property, which prohibits a task from accessing objects allocated by another task that might execute in parallel. Leveraging the disentanglement property, we present a garbage collection technique that can operate on any subtree in the memory hierarchy concurrently as other tasks (and/or other collections) proceed in parallel. We prove the safety of this collector by formalizing it in the context of our parallel calculus. In addition, we describe how the proposed techniques can be implemented on modern shared-memory machines and present a prototype implementation as an extension to MLton, a high-performance compiler for the Standard ML language. Finally, we evaluate the performance of this implementation on a number of parallel benchmarks.    

Ziv Scully                                        
A Program Optimization for Automatic Database Result Caching    
As Web applications scale up, they face challenges responding to a high volume of requests. A common aid is caching database results in the application's memory space, but the resulting modifications can be tricky to implement correctly and make the program less maintainable. In this paper, we present a compiler optimization that automatically adds sound SQL caching to Web applications coded in the Ur/Web language, with no modifications required to source code. We use a custom cache implementation that supports concurrent operations. Our experiments show that our optimization can double (or more) an application's throughput.

Tanmay Sinha                                        
Towards a Technology Design for Fostering Curiosity in Groupwork    
Curiosity is a vital socio-emotional skill in educational contexts. Yet, little is known about how social factors influence curiosity in small group learning. We argue that curiosity is evoked not only through individual, but also interpersonal activities, in particular peer interaction in group learning. We present what we believe to be the first theoretical framework that articulates an integrated psychological and social infrastructure of curiosity based on literature spanning psychology, learning sciences and group dynamics, and empirical observation of small group science learning across different learning environments. By making a tripartite distinction between observable behaviors, their individual and interpersonal functions and corresponding strategies to support those functions, we lay a foundation for a computational model of curiosity that is capable of guiding the design of learning technology, to recognize and evoke curiosity in small group learning. The underlying rationale is applicable more generally for modeling and developing computer support for other socio-emotional skills.

David Wajc                                        
A Faster Distributed Radio Broadcast Primitive    
We present a faster distributed broadcasting primitive  for the classical radio network model.     The most basic distributed radio network broadcasting  primitive - called Decay - dates back to a PODC'87 result of Bar-Yehuda, Goldreich, and Itai. In any radio network with some informed source nodes, running Decay for O(d log n + log^2 n) rounds informs all nodes at most d hops away from a source with high probability. Since 1987 this primitive has been the most important building block for implementing many other functionalities in radio networks. The only improvements to this decades-old algorithm are slight variations due to [Czumaj, Rytter; FOCS'03] and [Kowalski and Pelc, PODC'03] which achieve the same functionality in O(d log (n/d) +log^2 n) rounds. To obtain a speedup from this, d and thus also the network diameter need to be near linear.     Our new distributed primitive spreads messages for d hops in O(d log n log log n /log d +log^{O(1)} n) rounds with high probability. This improves over Decay for any super-polylogarithmic d and achieves near-optimal O(d log log n) running time for d = n. This also makes progress on an open question of Peleg.    

Colin White                                        
Clustering Under Natural Stability Assumptions    
Clustering is a fundamental problem in machine learning with many applications. A common approach is to set up an objective function and find the best solution according to the objective. In this work, we consider the k-center clustering objective, in which the goal is to find k centers to minimize the maximum distance from a point to its closest center. In the worst-case, this problem has a tight approximation ratio, but in this work, we go beyond the worst case and provide strong positive results for k-center under a very natural input stability condition called alpha-perturbation resilience.

Goran Zuzic                                        
Optimizations on Large Distributed Networks    
I am presenting a paper that creates a simple and practical framework for designing distributed algorithms on real-world networks. In particular, it allows for distributed Minimum Spanning Tree and Min-Cut Algorithms in near-optimal time.

Human Computer Interaction

Irene Alvarado                                        
Highlights from the Open Vis Conference    
Highlights and takeaways from the 2016 Open Vis Conference, an educational and forward-thinking conference on visualizing data on the web. Topics include data visualization, information design, data analysis, machine learning, D3, and more.

Anhong Guo                                        
Facade: Auto-generating Tactile Interfaces to Appliances    
Common appliances have shifted toward flat interface panels, making them inaccessible to blind people. Although blind people can label appliances with Braille stickers, doing so generally requires sighted assistance to identify the original functions and apply the labels. We introduce Facade - a crowdsourced fabrication pipeline to help blind people independently make physical interfaces accessible by adding a 3D printed augmentation of tactile buttons overlaying the original panel. Facade users capture a photo of the appliance with a readily available fiducial marker (a dollar bill) for recovering size information. This image is sent to multiple crowd workers, who work in parallel to quickly label and describe elements of the interface. Facade then generates a 3D model for a layer of tactile and pressable buttons that fits over the original controls. Finally, a home 3D printer or commercial service fabricates the layer, which is then aligned and attached to the interface by the blind person. We demonstrate the viability of Facade in a study with 11 blind participants.

Rushil Khurana                                        
CopyCat: Crowd-Enabled Electrical Muscle Stimulation (EMS) For Skill Transference    
CopyCat enables an application to control the movement of human limbs using muscle stimulation. The ability to control a human body via a software code can enrich the virtual reality experience by making the person 'feel' the same physiological responses in a VR movie or a game.

Meg Nidever, Clare Carroll, Catherine Chiodo, Adena Xin Lin, Jayanth Prathipati                  
Robin: Enabling Independence For Individuals With Cognitive Disabilities Using Voice Assistive Technology    
Individuals diagnosed with dementia are often most concerned about loss of independence, defined as the ability to continue living in one’s own home. Currently, assistive technologies that provide in-home cognitive support for those with degenerative brain diseases have not been widely adopted, and individuals often rely entirely on informal caregivers to aid them in planning and performing the daily activities that allow them to continue living independently. To help these individuals, we designed Robin, a conceptual context-aware assistive application that supports independent living for users with cognitive impairments by providing temporally and physically appropriate audio prompting for the routine tasks that are most important for health outcomes and life satisfaction. We designed conceptual application to work with existing technology (Amazon Alexa platform), and we are finalists for CHI's Student Design Competition 2017.

Joseph Seering                                        
Online Community Self-Regulation Techniques    
In this project we explore sociotechnical strategies used by online communities to integrate newcomers and deal with unwanted behaviors. This builds on previous research that identifies impact of indirect deterrence effects and differential impact of imitation effects based on levels of user status and authority. Through this work, we provide a theoretical framework for existing approaches to moderation online and to inform future designs by grounding them in an understanding of user strategies and preferences.

Alexandra To                                        
Treehouse Dreams: A Game-Based Method for Eliciting Interview Data from Children    
Treehouse Dreams is a card game that can simultaneously function as an ice breaker for children engaged in group activities, and as a research tool for interviewing children. The game, which asks players to collectively imagine a fantasy treehouse, incorporates game design techniques that encourage openness and sharing. The game is also easily customizable for a variety of research agendas. We describe a case study across two different lab environments, in which Treehouse Dreams was effectively used both to introduce young participants to the study environment and to draw out relevant information for research. 


Institute for Software Research

Kenneth Joseph                                        
Relating Semantic Similarity and Semantic Association to How Humans Label Other People    
Computational linguists have long relied on a distinction between semantic similarity and semantic association to explain and evaluate what is being learned by NLP models. In the present work, we take these same concepts and explore how they apply to an entirely different question - how individuals label other people. Leveraging survey data made public by NLP researchers, we develop our own survey to connect semantic similarity and semantic association to the process by which humans label other people. The result is a set of in- sights applicable to how we think of semantic similarity as NLP researchers and a new way of leveraging NLP models of semantic similarity and association as researchers of social science.

Sumeet Kumar                                         
Multimedia based Sentiment Analysis Dataset    
There has been a steady increase in multimedia based social interactions, but still most existing sentiment analysis tools are text-based. To create a multimedia based sentiment analyzer, we need a benchmark dataset. The dataset is essential to train, evaluate and optimize various methods of sentiment analysis. Since there is no known publicly available multimedia based sentiment dataset, we created a Tweets based gold standard dataset. The dataset contains sentiment labels for text tweets, image tweets and combined tweets (text + image). We expect that such a dataset will pave the way to create a multi-language multimedia based sentiment analysis tool.

Hemank Lamba                                        
Maximizing the Spread of Influence by Deadline    
Influence maximization has found applications in various fields such as sensor placement, viral marketing, controlling rumor outbreak, etc. In this paper, we propose a targeted approach to influence maximization in polarized networks i.e. networks where we already know or can predict node’s opinion about a product or topic. The goal is to find a set of individuals to target, such that positive opinion about a specific topic or the product to be launched is maximized.  Another key aspect that is present in most of the existing viral marketing algorithms is that they do not take into account the timeliness of the product adoption. In this paper, we present a framework where we infer the polarity, activity levels of the users, and then select seeds to launch viral marketing campaigns such that positive influence about the product is maximized by the given deadline.

Pardis Emami Naeini, Sruti Bhagavatula, Hana Habib, Martin Degeling                        
Privacy Expectations and Preferences in an IoT World    
With the rapid deployment of Internet of Things (IoT) technologies and the variety of ways in which they collect and use our data, there is a need for transparency, control, and new tools to ensure that people’s privacy requirements are met. To develop these tools, it is important to better understand how people feel about the associated privacy implications of IoT and the situations in which they want to be notified about data collection. We report on a large-scale vignette study with 1,007 participants focusing on people’s privacy expectations and preferences as they pertain to a collection of 380 data collection and use scenarios designed to capture a representative cross-section of IoT contexts. The scenarios vary according to eight factors, including the type of data collected (e.g., location, biometrics, temperature), how the data is used (e.g., whether it is shared and for what purpose), and other attributes, such as retention period.


Lane Center for Computational Biology

Brad Solomon                                        
Fast Search Using Sequence Bloom Trees    
An enormous amount of RNA sequence data has been published worldwide. In aggregate, this data could be used to investigate genetic variation, and condition- and disease-specific gene function in ways the original depositors of the data did not anticipate. However, searching the entirety of such a database has not been possible in reasonable computational time. Here we introduce a novel indexing data structure, the Sequence Bloom Tree, to address this gap between data and analysis.

Marcus Thomas                                        
Understanding Biological Self-Assembly    
The ability of collections of molecules to spontaneously assemble into large functional complexes is central to virtually all cellular processes. Virus capsid assembly has long been an important model system for understanding complicated self-assembly in part because of their combination of large scales yet geometric simplicity. Despite many years of study, though, many fine details of capsid assembly trajectories remain elusive. We have previously shown that local rule based simulation methods in conjunction with bulk indirect experimental data can meaningfully constrain the space of possible assembly trajectories and allow inference of experimentally unobservable features of the real system. Such a hybrid approach is beneficial because it is not possible to accurately measure the protein interaction rates used to parameterize any realistic model with current technology. We advance this strategy for data-driven model inference in two directions. First, we extend our prior work to encompass small-angle x-ray scattering (SAXS) as a possibly richer experimental data source than the previously used static light scattering. We further show how to bring such methods from the capsid domain to the important problem of amyloid aggregation.  Our results suggest that the reduction of noise by averaging large numbers of simulations at each parameter set is one way to improve the learning task, though more specialized algorithms offer the most promise.   

Language Technologies Institute

Jing Chen                                        
An Empirical Study of Learning to Rank for Entity Search    
This work investigates the effectiveness of learning to rank for entity search. Entities are represented by multi-field documents. Field-based text similarity features are extracted. We introduce learning to rank to entity search in DBpedia, and show that learning to rank is as powerful in entity search as in document search.
              

Machine Learning 

Kirstin Early                                        
Test Time Feature Ordering with FOCUS: Interactive Predictions with Minimal User Burden
Predictive algorithms are a critical part of ubiquitous computing, enabling appropriate action on behalf of users. Supervised machine learning algorithms make predictions from a set of features (selected at training time). However, features needed at prediction time may be costly to collect. In addition, feature cost and value may change dynamically based on real-world context (e.g., battery life) and prediction context (features already known and their values). We contribute a framework for dynamically trading off feature cost against prediction quality at prediction time. Our approach is up to 45% less costly than competing approaches, with equivalent or better error rates.

Calvin McCarter                                        
Large-Scale Optimization Algorithms for Sparse Conditional Gaussian Graphical Models
We address the problem of scalable optimization for sparse conditional Gaussian graphical models. Conditional Gaussian graphical models generalize the well-known Gaussian graphical models to conditional distributions to model the output network influenced by conditioning input variables. While highly scalable optimization methods exist for sparse Gaussian graphical model estimation, state-of-the-art methods for conditional Gaussian graphical models are not efficient enough and more importantly, fail due to memory constraints for very large problems. We propose a new optimization procedure based on a Newton method that efficiently iterates over two sub-problems, leading to drastic improvement in computation time compared to the previous methods. We then extend our method to scale to large problems under memory constraints, using block coordinate descent to limit memory usage while achieving fast convergence. Using synthetic and genomic data, we show that our methods can solve problems with millions of variables and tens of billions of parameters to high accuracy on a single machine.

Mrinmaya Sachan                                        
Machines Hit the School
An important roadblock in building an AI system is that unlike humans, machines lack any basic understanding of the world. Before we contemplate how we can achieve machine understanding, an interesting question to ask is: how does one define machine understanding? And how would we even measure machine understanding? Researchers often invoke the Turing test to this end (a machine attains human level intelligence if its responses in a dialogue with a human are indistinguishable from those of another human (Turing, 1950)), but as Levesque (2010) pointed out, this definition has resulted in researchers focusing on the wrong task, namely, fooling humans, rather than achieving machine intelligence. Shoehorning research to meet the goal of appearing human-like is a red herring. Levesque (2010) instead suggests multiple choice standardized tests as a suitable replacement for the Turing test. In this presentation, I will describe an elaborate set of standardized tests that have been proposed by many researchers recently as a replacement to the Turing test. For some of them, I will describe our initial attempts to solve them.


Robotics Institute

Wen-Sheng Chu                                        
Deep Region and Multi-label Learning for Facial Action Unit Detection
Region learning (RL) and multi-label learning (ML) have recently attracted increasing attentions in the field of facial Action Unit (AU) detection. Knowing that AUs are active on sparse facial regions, RL aims to identify these regions for a better specificity. On the other hand, a strong statistical evidence of AU correlations suggests that ML is a natural way to model the detection task. In this paper, we propose Deep Region and Multi-label Learning (DRML),a unified deep network that simultaneously addresses these two problems. One crucial aspect in DRML is a novel region layer that uses feed-forward functions to induce important facial regions, forcing the learned weights to capture structural information of the face. Our region layer serves as an alternative design between locally connected layers (i.e., confined kernels to individual pixels) and conventional convolution layers (i.e., shared kernels across anentire image). Unlike previous studies that solve RL and MLalternately, DRML by construction addresses both prob-lems, allowing the two seemingly irrelevant problems to in-teract more directly. The complete network is end-to-endtrainable, and automatically learns representations robustto variations inherent within a local region. Experiments onBP4D and DISFA benchmarks show that DRML performsthe highest average F1-score and AUC within and acrossdatasets in comparison with alternative methods.

Achal Dave, Jinyan Wang                                        
Predictive Corrective Networks for Action Detection
While deep feature-learning has revolutionized techniques for static-image understanding, the same does not quite hold for video processing. One reason may be that architectures and optimization techniques used for video are largely based off those for static images. In this work, we rethink both the underlying network architecture and the stochastic learning paradigm for temporal data. To do so, we draw inspiration from classic theory on linear dynamic systems for modeling time series. By extending such models to include nonlinear mappings, we derive a series of novel recurrent neural networks that sequentially make top-down {\bf predictions} about the future and then {\bf correct} those predictions with bottom-up observations.

Wenhao Luo                                        
Distributed Knowledge Leader Selection for Multi-Robot Environmental Sampling Under Bandwidth Constraints
In this project, we study the knowledge leader selection problem in multi-robot environmental sampling, where the goal is to select a subset of robots with a given cardinality due to the bandwidth constraint on human-swarm communication links, and those selected subset of robots can transmit the most informative task-specific data for the human. We prove that the knowledge leader selection is a submodular function maximization problem under explicit conditions and present a novel distributed submodular optimization algorithm that has the same approximation guarantees as the centralized greedy algorithm. The effectiveness of our approach is demonstrated using numerical simulations.

Nick Rhinehart                                        
Learning Action Maps of Large Environments via First-Person Vision
When people observe and interact with physical spaces, they are able to associate functionality to regions in the environment. Our goal is to automate dense functional understanding of large spaces by leveraging sparse activity demonstrations recorded from an ego-centric viewpoint. The method we describe enables functionality estimation in large scenes where people have behaved, as well as novel scenes where no behaviors are observed. Our method learns and predicts "Action Maps", which encode the ability for a user to perform activities at various locations. With the usage of an egocentric camera to observe human activities, our method scales with the size of the scene without the need for mounting multiple static surveillance cameras and is well-suited to the task of observing activities up-close. We demonstrate that by capturing appearance-based attributes of the environment and associating these attributes with activity demonstrations, our proposed mathematical framework allows for the prediction of Action Maps in new environments. Additionally, we offer a preliminary glance of the applicability of Action Maps by demonstrating a proof-of-concept application in which they are used in concert with activity detections to perform localization.

Arun Srivatsan Rangaprasad                                        
Complementary Model Update: A Method for Simultaneous Registration and Stiffness Mapping in Flexible Environments
Registering a surgical tool to an a priori model of  the environment is an important first step in computer-aided  surgery. In this paper we present an approach for simultaneous  registration and stiffness mapping using blind exploration of  flexible environments. During contact-based exploration of flexible  environments, the physical interaction with the environment  can induce local deformation, leading to erroneous registration  if not accounted for. To overcome this issue, a new registration  method called complementary model update (CMU), is introduced.  By incorporating measurements of the contact force,  and contact location, we minimize a unique objective function  to cancel out the effect of local deformation. We are thus able  to acquire the necessary registration parameters using both  geometry and stiffness information. The proposed CMU method  is evaluated in simulation and using experimental data obtained  by probing silicone models and an ex vivo organ.

Nitish Thatte                                        
Balance Recovery Control for Amputees using Powered Leg Prostheses
Currently in the US there are roughly 600,000 lower-limb amputees. These amputees fall as often as the elderly, resulting in physical injury and a fear of physical activity. To combat these issues, we are investigating a control strategy for robotic knee and ankle prostheses that is based on models of leg muscles and hypothesized reflexes. We seek to evaluate the ability of this control strategy to help amputees recover from disturbances. To do this, we plan to perform experiments with custom robotic prosthesis hardware and a mechanical disturbance device.

Marynel Vazquez                                        
Methods for Studying Group Interactions in HRI
In recent years, we have conducted several Human-Robot Interaction (HRI) experiments with small groups of people. To do so, we developed four different protocols to investigate human spatial behavior or trust in robots. We now look back at these efforts and highlight the opportunities and challenges of each experimental method. We also describe various group phenomena that we observed during the interactions. By sharing our experience, we hope to inform the community of the lessons that we learned in HRI and emphasize the importance of studying group interactions to enable robots to operate in public human environments.


Accounting

Eunhee Kim                                        
The Market for Reputation: CEO Turnover and Firm Performance
I propose a multiperiod matching model to explore CEO turnover-performance sensitivity, a relationship widely considered to be economically small. To determine whether to replace a CEO, each firm trades off atching efficiency and the incumbent CEO’s career concerns.  This trade-off can produce an equilibrium in which average performing CEOs are replaced and poor performing CEOs are retained. While average performing CEOs are perceived as better than poor performing CEOs, career concerns are more severe for average performing CEOs, thus making them difficult to incentivize. The model explains that turnover at the middle and retention at the bottom jointly weaken the observed association between turnover and performance, suggesting that the previously observed empirical relationship is essentially an outcome of the market for CEOs.

Economics

Eungsik Kim                                        
Accounting for a Positive Correlation between Pension and Consumption Taxes
We attempt to account for a puzzling feature of a positive correlation between pension level and consumption tax rate observed in the OECD data. First, using a standard overlapping generations model with lifetime uncertainty, we can determine optimal policy mix of taxes and pension, but we find that optimal tax and pension policy combinations cannot account for the data. Second, to resolve this puzzle, we consider welfare states where pension level is higher than the optimal level due to external and/or institutional reasons. In this setting, our analysis of optimal tax mix demonstrates that strengthening consumption (relative to income taxation) can improve welfare, i.e., accounting for the proposed puzzle. Third, we also find that population aging strengthens the role of consumption taxation, reinforcing our main findings. Finally, our results lend support to recent pension reforms: when expanding welfare benefits, most countries tend to resort to consumption tax financing.

Financial Economics

Camilo Botia Chaparro                                        
How Much Information Is Too Much Information? Lagged Disclosure, Bank Runs, and Risk Taking
I study the effects of disclosing financial information on the occurrence of bank runs and on management risk-taking activities.  The main trade-off is between the risk of bank runs, which increases with a disclosure delay, and managerial incentives for risk taking, which runs discipline.  I find that the main policy consideration is the growth rate of bank assets.  If bank assets grow sufficiently slowly, then the optimal policy is to disclose with a lag, in order to  balance managerial risk taking and creditors’ coordination problems.  When bank assets have high-growth rates, a disclosure lag increases the occurrence of runs and decreases bank value.    

     

Marketing

Michael Bernardi, Geoff Honda, Derek Jackman, Carol Marques                    
Reaching Out LGBT MBA Conference Overview
An overview of the 2016 Reaching Out LGBT MBA & Business Graduate Conference, which provides lesbian, gay, bisexual, transgender and queer young professional leaders from around the world the opportunity to network, learn, and improve their skills.
    
Francisco Cisternas Vera                                        
The Impact of New Technologies on Firm-consumer Relationships
I study the Mobile banking innovation. The proportion of US bank customers using Mobile banking has grown from 29% in 2012 to 43% in 2015. This channel of interaction is likely to keep growing due to a further increase in the adoption of smart phones, improvement in quality of Mobile banking apps and channel awareness. As a consequence some banks have reported that they may reduce their number branches by half over the next decade. The adoption of mobile banking displaces many banking functions performed through other channels like: automated teller machines (ATM), telephone banking, and online banking. Using geo-coded transaction data from a large consumer bank, a dynamic structural model to represent consumers’ preferences is developed for online and physical channels. In this way changes in banking behavior due to variation in the branch network structure as well as the introduction of the mobile channel are considered. This model is used to predict the timing and type of transactions across channels. The knowledge gained with the demand model is then used to design an optimal branch network in terms of capacities, amenities, location, and number of branches. Counterfactuals allow to evaluate different levels of channel adoption, and considers its effect on banking transactions, and more important, the impact on customer loyalty.  The model shows all channels remain relevant and, moreover, we found a strong complementarity between physical and digital worlds. Therefore instead of reducing the number of physical branches, banks should aim to adjust current branch capacities, specializing on transactions that cannot be served with digital channels. In conclusion digital channels will diminish but never replace physical channels and they will be redesigned correspondingly. Is important to note that this is the first time in the banking industry that substitution of branches for digital channels are formally considered in a tool to support branch network design for the middle and long term.

Julian Givi                                        
Why Certain Gifts Are Great to Give But Not to Get: A Framework for Understanding Errors in Gift Giving
We propose that when evaluating the quality of a gift, givers primarily focus on the ‘moment of the exchange,’ whereas recipients instead mostly focus on how valuable a gift will be throughout their ownership of it. This theory can account for many of the errors documented in the gift giving literature, make novel predictions about future errors in gift giving, and offer insight into how givers can give better gifts.

Yijin Kim                                        
How Airbnb Affects Local Rental Market
This research project studies the impact of Airbnb on the rental housing market. We construct a structural model of property owners’ choice between long-term rental and Airbnb and their decision on the number of days to list their properties on Airbnb if and when they choose Airbnb. The model can be estimated using the American Housing Survey (AHS) data and Airbnb data. We illustrate how the parameter estimates from the model can be used to assess the impact of Airbnb on rental housing supply and affordability. We can also use the parameter estimates to investigate how property owners would respond to various policy interventions, such as limiting the number of days that a property can be rented out, imposing taxes on short-term rentals, and rent controls.

Operation Management & Manufacturing

Gerdus Benade                                        
Preference Elicitation for Participatory Budgeting
Participatory budgeting enables the allocation of public funds by collecting and aggregating individual preferences; it has already had a sizable real-world impact. But making the most of this new paradigm requires a rethinking of some of the basics of computational social choice, including the very way in which individuals express their preferences. We analytically compare four preference elicitation methods --- knapsack votes, rankings by value or value for money, and threshold approval votes ---  and find that threshold approval votes are qualitatively superior. This conclusion is supported by experiments using data from real participatory budgeting elections.

Nam Ho-Nguyen                                        
Dynamic Data-Driven Estimation of Non-Parametric Choice Models
We study non-parametric choice models to predict consumer choice behaviour from observational data. Estimating a non-parametric model requires optimizing a factorial number of variables, which is intractable even for moderate-scale problems. We present a general framework based on convex conjugacy, saddle point duality and online convex optimization to estimate a non-parametric choice model. Our method enjoys provable convergence guarantees and extends naturally to the dynamic case where we update our dataset during estimation. However, our method encounters a combinatorial subproblem inherent to the non-parametric approach. We identify structural conditions under which the subproblem can be solved efficiently.

Organizational Behavior and Theory

Erin Fahrenkopf                                        
A Nudge to Enter: When Prior Organizational Experiences Spur Entrepreneurial Activity
Using an experiment in which participants gain experience as specialists or generalists, the study provides causal evidence on the effect of specialization on potential entrepreneurs’ entry decisions. The expected findings further our understanding of the antecedents of entrepreneurship by showing that increasing the degree of general knowledge in individuals’ backgrounds increases their likelihood of taking on entrepreneurial endeavors.

Alessandro Iorio                                        
Divide and Rule: A Network Assessment of an Italian Anti-Interlocking Law
On December 6, 2011, in an attempt to limit the power of the financial sector, the Italian government issued an anti-interlocking law preventing financial firms from sharing board members with one another. To evaluate the impact of the law, we collected longitudinal data on board interlocks among all the financial and non-financial firms listed in the Italian stock market from 1998 to 2015. We found strong evidence that the anti-interlocking law impaired the level of cohesiveness among the financial corporate elite.

Yeonjeong Kim                                        
Uncovering Moral Character via Interview Questions
This study investigates whether groups of naïve judges can detect peoples’ moral character in interview settings. We conducted studies in which we elicited moral character information from strangers via behavioral interview questions designed to covertly reveal people’s moral character through their spontaneous written responses. We found that six judges were enough to reliably estimate targets’ moral character. Across three studies, the average of judges’ moral character evaluations predicted targets’ unethical behaviors, demonstrating the validity of these interview-based methods.

Anna Mayo                                        
Field Evidence for Collective Intelligence in Business Unit Performance.
Using panel data from sales units in a national financial institution, we find evidence of collective intelligence, operationalized as a consistent ability to capitalize on resources, which explains sales-unit performance above and beyond other individual, sales-unit, and market features. We also explore the role of coordination among sales unit members.

Amanda Weirup                                        
Will You Do Me a Favor? Responding to Favor Requests in the Workplace
Decisions about whether to perform favors for colleagues, supervisors, and subordinates are an important issue that faces all working professionals as they try to balance the many divergent demands on their time. This work focuses on how individuals make decisions regarding whether to agree to favors, defined as “discretionary, prosocial behavior that is performed in response to a specific, explicit request from one person to another.” We consider favor requests from the perspective of the performer (as opposed to the requestor) to identify the motivations and emotions that influence responses to favor requests and consider how favor decision making differs across individuals and situations.





Entertainment Technology Center

Alejandra Soto, Antonin Fusco                                        
Learner’s Behavior Professional Development Card Deck
The card deck is a result of a research and discovery project hosted by Carnegie Mellon University’s Entertainment Technology center during the spring semester of 2016. The students working on the project were tasked with helping the Children’s Museum of Pittsburgh’s makerspace, the MAKESHOP©, promote family unit making.      Through collaboration with the makerspace’s educators, it was found that a common language on how to design and interact with guests of varied ages was needed. The cards provide this by helping users with brainstorming ideas, communicating goals, and identifying the needs of different types of guests based on their personalities.  

Jun Wang                                        
AR and VR User Experience Design Practice
This project summarizes my discoveries of UI/UX design practice for emerging new platforms such as AR and VR.

Integrated Innovation Institute

Joelle El Hayek                                        
GHC - Empowering Women in Technology
Highlights of GHC, which focused on empowering women in technology through:  - an international career fair  - tech talks  - personal and professional development workshops   - mentorship opportunities.