Seminar on Computational Learning and Adaptation


  Revising Engineering Models: Combining Computational Discovery with Knowledge

Stephen Bay
Institute for the Study of Learning and Expertise, and
Computational Learning Laboratory, CSLI
sbay@apres.stanford.edu
http://newatlantis.isle.org/~sbay

Developing mathematical models that represent physical devices is a difficult and time consuming task. In this talk, I present a hybrid approach to modeling that combines machine learning methods with knowledge from a human domain expert. Specifically, I propose a system for automatically revising an initial model provided by an expert with an equation discovery program that is tightly constrained by domain knowledge. I apply this system to learning an improved model of a battery on the International Space Station from telemetry data. The results suggest that this hybrid approach can reduce model development time and improve model quality.

This talk describes joint work with Dan Shapiro and Pat Langley.



Date: Thursday, May 23

Time: 4:15-5:30PM

Place: Cordura 100


Return to the seminar schedule