Results 1 -
3 of
3
Evaluating Evaluations: A Comparative Study of Metrics for Comparing Learning Performances Investigators
"... One of the challenges in comparing learning performance across multiple conditions is to develop an appropriate evaluation. We identify four candidate metrics and apply them in an empirical example, providing contrast to differences between the metrics. We propose a set of criteria that learning com ..."
Abstract
- Add to MetaCart
One of the challenges in comparing learning performance across multiple conditions is to develop an appropriate evaluation. We identify four candidate metrics and apply them in an empirical example, providing contrast to differences between the metrics. We propose a set of criteria that learning comparison metrics should satisfy, evaluate the four metrics using the
Modelling a RTS Planning Domain with Cost Conversion and Rewards
"... Abstract. Game playing has always been a good testbed for AI research. In particular, real time strategy (RTS) games are characterized by being complex environments which offer multiple opportunities for AI researchers. In this work, we use AI planning and apply it to a RTS game. More specifically, ..."
Abstract
- Add to MetaCart
Abstract. Game playing has always been a good testbed for AI research. In particular, real time strategy (RTS) games are characterized by being complex environments which offer multiple opportunities for AI researchers. In this work, we use AI planning and apply it to a RTS game. More specifically, we have defined a domain using PDDL2.1 that can be used by a computer player to compete in a RTS game setting using automated planning. The goal is to participate in the annual ORTS competition and we have focused on one of its tasks. We have used techniques such as cost to operator conversion and rewarded soft goals to overcome the difficulties of modelling this kind of tasks. 1
