Corpus ID: 53033974

Learning Classical Planning Strategies with Policy Gradient

@inproceedings{Gomoluch2019LearningCP,
  title={Learning Classical Planning Strategies with Policy Gradient},
  author={Pawel Gomoluch and Dalal Alrajeh and A. Russo},
  booktitle={ICAPS},
  year={2019}
}
  • Pawel Gomoluch, Dalal Alrajeh, A. Russo
  • Published in ICAPS 2019
  • Computer Science
  • A common paradigm in classical planning is heuristic forward search. Forward search planners often rely on relatively simple best-first search algorithm, which remains fixed throughout the search process. In this paper, we introduce a novel search framework capable of alternating between several forward search approaches while solving a particular planning problem. Selection of the approach is performed using a trainable stochastic policy. This enables tailoring the search strategy to a… CONTINUE READING

    Figures, Tables, and Topics from this paper.

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 28 REFERENCES
    The LAMA Planner: Guiding Cost-Based Anytime Planning with Landmarks
    536
    The FF Planning System: Fast Plan Generation Through Heuristic Search
    1978
    Learning to Rank for Synthesizing Planning Heuristics
    11
    Marvin: A Heuristic Search Planner with Online Macro-Action Learning
    116
    A review of machine learning for automated planning
    62
    Best-First Width Search: Exploration and Exploitation in Classical Planning
    38
    The Fast Downward Planning System
    1202