This research addresses the formation of new concepts and their corresponding ontology in a multiagent system where individual autonomous agents try to learn new concepts by consu...
PAC-MDP algorithms approach the exploration-exploitation problem of reinforcement learning agents in an effective way which guarantees that with high probability, the algorithm pe...
This paper documents progress to date on a research project, the goal of which is wartime event prediction. The paper describes the operational concept, the datamining environment...
— The first-person-shooter Quake II is used as a platform to test neuro-visual control and retina input layouts. Agents are trained to shoot a moving enemy as quickly as possibl...
Human experience with interactive games will be enhanced if the software agents that play the game learn from their failures. Techniques such as reinforcement learning provide one...