Greg Ver Steeg
Greg Ver Steeg, Ph.D.
Research Assistant Professor, University of Southern CaliforniaMy current endeavor is to develop an information-theoretic framework that enables fast learning of succinct representations that capture complex phenomena. The first few steps in this program are here:
- Prototype with sample applications (NIPS-14) and code
- Information-theoretic foundation for learning deep representations (AISTATS-15)
- How information in complex systems can be extracted incrementally using the "information sieve" method
As intelligent beings in a complex world, we are constantly faced with the challenge of making sense of it all through simplification and abstraction. Our abstract representations of the world seem to be most useful when they "carve nature at its joints" or "perceive and bring together in one idea the scattered particulars" (as Socrates puts it). The motivation of this research program is to formalize this idea in a practical and general way. Information theory gives us a way to measure how informative a representation is about the world. One bonus is that we find this information about the world can be decomposed modularly and hierarchically. A second bonus is that we are able to efficiently search for representations that are as informative as possible about some set of observations. We have been using these ideas to search for abstract representations that help us understand complex data like neurophysiology of Alzheimer's patients, gene expression of cancer patients, and human behavior. Ultimately, I hope these ideas will lead to a better understanding of intelligence.
Other ongoing research efforts
|Email:||gregv at isi.edu|