Impact of Censored Sampling on the Performance of Restart Strategies

Algorithm selection, algorithm portfolios, and randomized restarts, can profit from a probabilistic model of algorithm run-time, to be estimated from data gathered by solving a set of experiments. Censored sampling offers a principled way of reducing this initial training time. We study the trade-off between training time and model precision by varying the censoring threshold, and analyzing the consequent impact on the performance of an optimal restart strategy, based on an estimated model of runtime distribution. We present experiments with a SAT solver on a graph-coloring benchmark. Due to the “heavy-tailed” runtime distribution, a modest censoring can already reduce training time by a few orders of magnitudes. The nature of the optimization process underlying the restart strategy renders its performance surprisingly robust, also to more aggressive censoring.

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

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

[3]  Erkki Oja,et al.  Artificial Neural Networks: Formal Models and Their Applications - ICANN 2005, 15th International Conference, Warsaw, Poland, September 11-15, 2005, Proceedings, Part II , 2005, International Conference on Artificial Neural Networks.

[4]  Wayne B. Nelson,et al.  Applied Life Data Analysis: Nelson/Applied Life Data Analysis , 2005 .

[5]  Thomas Stützle,et al.  SATLIB: An Online Resource for Research on SAT , 2000 .

[6]  Holger H. Hoos,et al.  On the Run-time Behaviour of Stochastic Local Search Algorithms for SAT , 1999, AAAI/IAAI.

[7]  Irina Rish,et al.  Summarizing CSP Hardness with Continuous Probability Distributions , 1997, AAAI/IAAI.

[8]  William J. Reed,et al.  The Double Pareto-Lognormal Distribution—A New Parametric Model for Size Distributions , 2004, WWW 2001.

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

[10]  Shlomo Zilberstein,et al.  Monitoring and control of anytime algorithms: A dynamic programming approach , 2001, Artif. Intell..

[11]  Holger H. Hoos,et al.  A mixture-model for the behaviour of SLS algorithms for SAT , 2002, AAAI/IAAI.

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

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

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

[15]  Jürgen Schmidhuber,et al.  A Neural Network Model for Inter-problem Adaptive Online Time Allocation , 2005, ICANN.

[16]  David Zuckerman,et al.  Optimal Speedup of Las Vegas Algorithms , 1993, Inf. Process. Lett..

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

[18]  Christian Bessiere,et al.  Statistical Regimes Across Constrainedness Regions , 2004, Constraints.

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

[20]  Leonidas J. Guibas,et al.  A method for obtaining randomized algorithms with small tail probabilities , 1996, Algorithmica.

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

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

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

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

[25]  Bart Selman,et al.  Boosting Combinatorial Search Through Randomization , 1998, AAAI/IAAI.

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

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

[28]  Thomas Stützle,et al.  Local Search Algorithms for SAT: An Empirical Evaluation , 2000, Journal of Automated Reasoning.