  Sequential decision tasks appear in many practical real-world problems including control, resource allocation, and routing.  Such tasks can be characterized by the following scenario: An agent observes a state of a dynamic system and selects from a finite set of actions.  The system then enters a new state upon which the agent must select another action.  The system may return a payoff for each decision made or for a sequence of decisions.  The objective is to select the sequence of actions that return the highest total or cumulative payoff.  In my research, I evolve Neural Networks with Genetic Algorithms to learn and perform sequential decision tasks.  I am particularly interested in tasks where problem-specific knowledge is currently unavailable or costly to obtain.  Some domains that I have studied include game playing, intelligent control, and constraint satisfaction.  For more information, see my <a href="http://www.cs.utexas.edu/users/moriarty/moriarty-papers.html"> list of publications</a>. <p>
