

* do a theoretical and empirical comparison of statistical vs. neural
techniques for pattern classification and/or function approximation


some student projects from Fall `91 course


Jim Blythe
  something to do with the Granger & Lynch piriform cortex model

Rich Caruana
  effect of network architecture on generalization

John Cheng
  extension of Pollack's dynamical recognizer results

Scott Reilly
  Kohonen-like networks for phoneme recognition

Xu-Mei Wang
  cognitive phonology using deterministic Boltzmann machines

Mark Wheeler
  using reinforcement learning to play Mille Bornes

Dan Zhu
  something to do with life/death decisions in the game of Go
