Learning Restart Strategies

Restart strategies are commonly used for minimizing the computational cost of randomized algorithms, but require prior knowledge of the run-time distribution in order to be effective. We propose a portfolio of two strategies, one fixed, with a provable bound on performance, the other based on a model of run-time distribution, updated as the two strategies are run on a sequence of problems. Computational resources are allocated probabilistically to the two strategies, based on their performances, using a well-known K-armed bandit problem solver. We present bounds on the performance of the resulting technique, and experiments with a satisfiability problem solver, showing rapid convergence to a near-optimal execution time.

[1]  Aaas News,et al.  Book Reviews , 1893, Buffalo Medical and Surgical Journal.

[2]  Huayue Wu,et al.  Randomization and Restart Strategies , 2006 .

[3]  Mauricio Marín,et al.  Collecting data about moving objects in a distributed environment , 2004, XXIV International Conference of the Chilean Computer Science Society.

[4]  Tad Hogg,et al.  The Hardest Constraint Problems: A Double Phase Transition , 1994, Artif. Intell..

[5]  Jürgen Schmidhuber,et al.  Impact of Censored Sampling on the Performance of Restart Strategies , 2006, CP.

[6]  Thomas G. Dietterich.  What is machine learning? , 2020, Archives of Disease in Childhood.

[7]  Michail G. Lagoudakis,et al.  Algorithm Selection using Reinforcement Learning , 2000, ICML.

[8]  Toby Walsh,et al.  Morphing: Combining Structure and Randomness , 1999, AAAI/IAAI.

[9]  David Zuckerman,et al.  Optimal speedup of Las Vegas algorithms , 1993, [1993] The 2nd Israel Symposium on Theory and Computing Systems.

[10]  Stephen F. Smith,et al.  The Max K-Armed Bandit: A New Model of Exploration Applied to Search Heuristic Selection , 2005, AAAI.

[11]  Tad Hogg,et al.  An Economics Approach to Hard Computational Problems , 1997, Science.

[12]  Marek Petrik,et al.  Statistically Optimal Combination of Algorithms , 2004 .

[13]  Jürgen Schmidhuber,et al.  Dynamic Algorithm Portfolios , 2006, AI&M.

[14]  Nicolò Cesa-Bianchi,et al.  Gambling in a rigged casino: The adversarial multi-armed bandit problem , 1995, Proceedings of IEEE 36th Annual Foundations of Computer Science.

[15]  Katinka Wolter,et al.  Analysis and algorithms for restart , 2004 .

[16]  Jürgen Schmidhuber,et al.  Learning dynamic algorithm portfolios , 2006, Annals of Mathematics and Artificial Intelligence.

[17]  Nils J. Nilsson,et al.  Artificial Intelligence , 1974, IFIP Congress.

[18]  Katinka Wolter,et al.  Analysis and algorithms for restart , 2004, First International Conference on the Quantitative Evaluation of Systems, 2004. QEST 2004. Proceedings..

[19]  Bart Selman,et al.  Algorithm portfolios , 2001, Artif. Intell..

[20]  Mark S. Boddy,et al.  Deliberation Scheduling for Problem Solving in Time-Constrained Environments , 1994, Artif. Intell..

[21]  E. Kaplan,et al.  Nonparametric Estimation from Incomplete Observations , 1958 .

[22]  Wayne Nelson,et al.  Applied life data analysis , 1983 .

[23]  Ricardo Vilalta,et al.  Introduction to the Special Issue on Meta-Learning , 2004, Machine Learning.

[24]  H. Robbins.  Some aspects of the sequential design of experiments , 1952 .

[25]  John R. Rice,et al.  The Algorithm Selection Problem , 1976, Adv. Comput..

[26]  Chu Min Li,et al.  Heuristics Based on Unit Propagation for Satisfiability Problems , 1997, IJCAI.

[27]  Stephen F. Smith,et al.  An Asymptotically Optimal Algorithm for the Max k-Armed Bandit Problem , 2006, AAAI.

[28]  Bart Selman,et al.  Heavy-Tailed Phenomena in Satisfiability and Constraint Satisfaction Problems , 2000, Journal of Automated Reasoning.

[29]  Eric Horvitz,et al.  Dynamic restart policies , 2002, AAAI/IAAI.