Output details
11 - Computer Science and Informatics
University of Strathclyde
Evolutionary-based learning of generalised policies for AI planning domains
<22>The significance of this paper is that it isolates a large sub-class of planning domains that can be solved using a rule-based policy and further shows that evolutionary machine learning can create solvers that outperform the previous state-of-the-art solutions. Further evidence of significance is that the technique has been used by other researchers in solving hard sequential decision problems (e.g. by Sipper et al.). Finally this paper was published in the best conference in evolutionary computation (GECCO) and was cited by Pappa et al. as an exemplar of applying evolutionary machine learning to the problem of domain independent planning.