Current MS Students/Rick Strom
From CSWiki
Contents |
[edit] Abstract
New research into neural networks for artificial intelligence favors a break from traditional feed-forward networks and a move towards highly recurrent, complex networks. Questions have also been raised about the importance of psychology and emotion on intelligence. I am addressing a sub-problem of this, namely, the impact of emotion on learning.I am presenting a reccurrent neural net class which incorporates these ideas. The class includes a component that can alter the degree of Hebbian learning taking place according to a measure of positive or negative reinforcement from the system. Additionally, I am presenting a simulation arena application in which a wide variety of challenges can be given to the AI to determine their success.
