Lab Home | Phone | Search
Center for Nonlinear Studies  Center for Nonlinear Studies
 Colloquia Archive 
 Postdoc Seminars Archive 
 Quantum Lunch 
 Quantum Lunch Archive 
 CMS Colloquia 
 Q-Mat Seminars 
 Q-Mat Seminars Archive 
 P/T Colloquia 
 Kac Lectures 
 Kac Fellows 
 Dist. Quant. Lecture 
 Ulam Scholar 
 CNLS Fellowship Application 
 Student Program 
 Past Visitors 
 History of CNLS 
 Maps, Directions 
 CNLS Office 
Thursday, July 30, 2009
12:30 PM - 1:15 PM
CNLS Conference Room (TA-3, Bldg 1690)


Synthetic Visual Cognition Student Seminar: On the Fokker-Planck Analysis of Neural Network Learning Rules

David Lyttle

A number of simple learning rules have been proposed and implemented in artificial neural networks. To date, most attempts to understand the long-term dynamics of these rules have been based around simulations and experiments. An alternative approach is to try and understand the behavior of these rules from an analytical standpoint. Here we attempt to do so for a few simple cases using a Fokker-Planck formalism, and show that for some rules, the analysis yields meaningful results, while for others the approach breaks down or becomes very difficult. Despite these difficulties we are still able to highlight basic differences between learning rules. Furthermore we can identify several potential avenues by which this research can be extended in the future.

Host: Garrett Kenyon