Learning DepthFirst Search A Unified Approach to Heuristic Search in Deterministic and Non.pdfVIP

Learning in Depth-First Search: A Unified Approach to Heuristic Search in Deterministic, Non-Deterministic, Probabilistic, and Game Tree Settings ´ Blai Bonet Hector Geffner ´ Departamento de Computacion ICREA Universitat Pompeu Fabra ´ ´ ´ Universidad Simon Bolıvar Paseo de Circunvalacion, 8 Caracas, Venezuela Barcelona, Spain bonet@ldc.usb.ve hector.geffner@upf.edu Abstract then the solution is found and reported, else the process restarts with the lower bound and the value function updated. Dynamic Programming provides a convenient and unified framework for studying many state models used in AI but no LDFS reduces to IDA * with Transposition Tables (Reinefeld algorithms for handling large spaces. Heuristic-search meth- Marsland 1994) over deterministic models, but solves also ods, on the other hand, can handle large spaces but lack a non-deterministic, probabilistic, and game tree models, over common foundation. In this work, we combine the bene- which a slight variation reduces to the state-of-the-art MTD fits of a general dynamic programming formulation with the algorithm (Pla