Output details
11 - Computer Science and Informatics
Imperial College London
Learning Revised Models For Planning In Adaptive Systems
<09> Model revision is extremely challenging. This paper demonstrates the benefits of integrating ILP, model checking and plan synthesis in adaptive systems for revising models which can be incomplete, inaccurate and/or out of date. The approach uses non-monotonic probabilistic rule learning (NoMPRoL) to find hypotheses that explain the observations and conditions under which they occur. Updated models are used to generate new plans. Research has led to continuing collaboration with the NII Japan (Prof. Honiden http://research.nii.ac.jp/~honiden/) and was presented at the flagship software engineering conference, ICSE: Acceptance 18.5%/461. Preliminary work was part of the keynote talk, ICSE 2012.