Artificial Intelligence (CS365)

Learning Heuristic Functions for 24-Puzzle

Mentor : Dr.Amitabha Mukerjee

GROUP S2
Vinit Kataria & N.V.Subba Rao

Abstract

Solving large state space problems within the given computational limits is an active area of research in the recent years and a lot of work is being done on finding better methods. In our project we study one such method, the Bootstrapping Procedure[1], used to solve 24 sliding tile puzzle and investigate the results associated with solving this large state space problem and also the Interleaving Procedure[1] for solving single problem instances. Experimental results indicate that the time taken for solving is reduced significantly with some reasonable sub-optimality in the solution cost.

PROPOSAL

PRESENTATION

REPORT

CODE(Random-walk & Interleaving)

References

  1. Shahab Jabbari Arfaee, Sandra Zilles, Robert C. Holte. Learning Heuristic Functions for Large State Spaces. In Journal of Artificial Intelligence,175:pages 2075-2098, 2011.
  2. Marco Ernandes and Marco Gori. Likely-admissible and sub-symbolic heuristics. In Proceedings of the 16th European Conference on Artificial Intelligence (ECAI 2004), pages 613-617, 2004.
  3. Jonathan Schaeffer, Ariel Felner, Mehdi Samadi. Learning from multiple heuristics. In Proceedings of 23rd AAAI Conference on Artificial Intelligence(2008), pages 357-362, 2008.
  4. Richard E. Korf and Ariel Felner. Disjoint pattern database heuristics. Artificial Intelligence, 134:9-22, 2002.
  5. Larry A. Rendell. A new basis for state-space learning systems and a successful implementation. Artificial Intelligence, 20:369-392, 1983.