October 24, 2019
Study on Remote Medical Robotics with Machine Learning Receives Research Grant
The research aims to develop machine learning technology for medical robotics during remote, life-saving surgeries.
Research on the integration of constraint reasoning and machine learning conducted by Willem-Jan van Hoeve, Carnegie Bosch Professor of Operations Research, received a “Formal Methods in the Field” research grant from the National Science Foundation (NSF).
The study, Embedding Constraint Reasoning in Machine Learning for Better Prediction and Decision-making, was performed by van Hoeve with collaborators Juan Pablo Wachs and Yexiang Xue, Associate Professor of Industrial Engineering and Assistant Professor of Computer Science at Purdue University, respectively. Wachs, in particular, has applied machine learning techniques to medical robotics in his research for several years, and heads the Intelligent Systems and Assistive Technologies Lab.
“Handling constraints within machine learning algorithms is a very important challenge in the field of artificial intelligence,” said van Hoeve. “And I think we have an interesting angle. We extend the neural network with an addition module that represents constraints generically and compactly.”
The researchers specifically aim to advance the use of data-driven machine learning and constraint reasoning technology in trauma care systems, enabling doctors to remotely perform life-saving surgery in hazardous situations such as on battlefields or after natural disasters such as hurricanes.
The FMitF program brings together researchers with backgrounds in theory and engineering to design new approaches with provable guarantees for challenging applications. The award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.
“What this stream of grants tries to do is to consider theoretical ideas, for example from computer science, mathematics and operations research, and develop and integrate them into actual application,” van Hoeve said.