Carnegie Mellon University

Joseph Ramsey

Joseph Ramsey

Special Faculty and Director of Research Computing

  • Porter Hall 126
  • 412.268.8063

Education

Ph.D. – 2001, Philosophy, University of California at San Diego
M.A. – 1995, Philosophy, University of California at San Diego

Departmental Research

I actively participate in or support a number of research projects in the department, in varying degrees. The two most active are Tetrad and AProS. Tetrad (with Clark Glymour, Richard Schienes, and Peter Spirtes) is a repository of Java code spanning causal search algorithms, estimation algorithms, clustering, updating, and so on, with a graphical user interface available freely over the internet. Much of this code I have programmed myself, and much of the rest of it I have been actively involved in developing. AProS (with Wilfried Sieg) is a project with several goals—to develop a natural deduction proof generator for sentential and first order logic, to develop an online course for logic, and to develop a proof tutor that uses the generator to provide advise to students for completing logic problems. Much of the code for the most recent version of this was developed by me and expounded on by later programmers. I have continued to participate in this project in the role of programmer, of supervisor, and of advisor.

There are many other projects in the department as well for which I provide ongoing support. To list a few, there is the Isabelle project under the direction of Jeremy Avigad, the Bernays and Carnap Translation projects under the direction of Steve Awodey, and the Causality Lab project under the direction of Richard Scheines. Depending on the project, I provide repository and server services, software advise and services, or (as in the case of Causality Lab) all of the above plus code. In addition to specific long-term projects, I often help individual faculty and graduate students with specific computational difficulties, although examples of this sort are too numerous to list.

Grant-Funded Research

  • The development of courseware in causal-statistical reasoning (FIPSE, Mellon Foundation). (link)
  • The development of on-board Rover software to detect the presence of carbonates in rocks (NASA)  [7, 8].
  • The development of software to simulate gene expression (NASA) [11, 12].
  • The analysis of causal effects in climate data (University of West Florida).
  • The analysis of causal effects in forest fire data (NASA).
  • The analysis of the use of OCT data to predict presence of glaucoma (University of Pittsburgh) [1].
  • The development and performance analysis of an online course in introductory logic (NSF). (link)
  • The development of software to detect chains of command in social situations [4] (NSA).
  • The development of algorithms to detect causal processes in the brain from fMRI data (NSF).

Algorithm Development

In recent years, I have helped to develop a number of algorithms, including the following:

  • Rockspec—Uses Tetrad-style techniques to detect the presence of minerals in rocks from their spectrographic signatures [7, 8].
  • CPC (“Conservative PC”) – Modifies the PC algorithm to do a more conservative orientation of edges [9].
  • MBFS (“Markov blanket fan search”) – Estimates the Markov blanket of variables in high-dimensional datasets [2, 3, 10].
  • Command – Estimates the command structure of a group of people from labeled, time-stamped documents [4].
  • LiNGAM Pattern – Adapts the LiNGAM algorithm (to search for acyclic linear causal structure with nongaussian errors) to the case where some errors are gaussian and others are nongaussian [5].

Infrastructure Support

In support of the department’s computational infrastructure, I am responsible for the purchase of computers in the department and for aspects of their maintenance that go beyond what other departments on campus will support. I maintain a server for the department that contains the code for most of our computational projects. In addition to this, I serve as the Director of Computing for the Laboratory for Symbolic and Educational Computing, under the direction of Wilfied Sieg and Teddy Seidenfeld.

Academic Career

2006 – present Special Faculty and Director of Research Computing, Department of Philosophy, Carnegie Mellon University, Pittsburgh, PA.

2002 - 2006 Director of Computing, Department of Philosophy, Carnegie Mellon University, Pittsburgh, PA.

2000-2002 Research Programmer and Director of Computing, Department of Philosophy, Carnegie Mellon University, Pittsburgh, PA.

1997 - 2000 Research Programmer, Department of Philosophy, Carnegie Mellon University, Pittsburgh, PA.

Selected Publications

[1] Z. Burgansky-Eliash, G. Wollstein, T. Chu, J. Ramsey, C. Glymour, R. Noecker, and J. Schuman (2005). Optical coherence tomography machine learning classifiers for glaucoma detection. Investigative Ophthamology & Visual Science, 2005.

[2] X. Bai, C. Glymour, R. Padman, J. Ramsey, P. Spirtes, and F. Wimberly. PCX:  Markov Blanket Classification for Large Data Sets with Few Cases (2004). Center for Automated Learning and Discovery, CMU-CALD-04-102, School of Computer Science, Carnegie Mellon University.

[3] Bai, X., R. Padman, J. Ramsey, and P. Spirtes. Tabu Search Enhanced Graphical Models for Classification in High Dimensions. INFORMS, forthcoming.

[4] D. A. Gerdes, C. Glymour, J. D. Ramsey. (2006). Who's calling? Deriving organization structure from communication records. In A. Kott (Ed)., Information Warfare and Organizational Decision-Making. Boston, MA: Artech House.

[5] P. O. Hoyer, A. Hyvärinen, R. Scheines, P. Spirtes, J. Ramsey, G. Lacerda, and S. Shimizu. Causal discovery of linear acyclic models with arbitrary distributions. Submitted, Uncertainty in Artificial Intelligence.

[6] G. Lacerda, P. Spirtes, J. Ramsey, and P. O. Hoyer. Discovering cyclic causal models by independent components analysis. Submitted, Uncertainty in Artificial Intelligence.

[7] J. Ramsey, P. Gazis, T. Roush, P. Spirtes and C. Glymour (2002). Automated remote sensing with near infrared reflectance spectra: carbonate recognition. Data Mining and Knowledge Discovery, v. 6, no. 3, 275-291.

[8] J. Ramsey (2001). Mixture and expertise in automatic causal discovery. Ph.D. diss., University of California at San Diego.

[9] J. Ramsey, P. Spirtes, and J. Zhang (2006). Adjacency-faithfulness and conservative causal inference. Proceedings of the 22nd Conference on Uncertainty in Artificial Intelligence, 401-408, Oregon, AUAI Press.

[10] J. Ramsey (2006). A PC-style Markov blanket search for high-dimensional datasets. Technical Report CMU-PHIL-177. Pittsburgh, PA: Carnegie Mellon University Philosophy Department.

[11] R. Scheines. and J. Ramsey (2001). Simulating genetic regulatory networks. Technical Report CMU-PHIL-24. Pittsburgh, PA: Philosophy Department, Carnegie Mellon University.

[12] F. Wimberly, T. Heiman, C. Glymour, and J. Ramsey (2003). Experiments on the accuracy of algorithms for inferring the structure of genetic regulatory networks from microarray expression levels. Proceedings of the Workshop on Learning Graphical Models in Computational Genomics, International Joint Conference on Artificial Intelligence, Acapulco.