Is perturbation an effective restart strategy?

Premature convergence can be detrimental to the performance of search methods, which is why many search algorithms include restart strategies to deal with it. While it is common to perturb the incumbent solution with diversification steps of various sizes with the hope that the search method will find a new basin of attraction leading to a better local optimum, it is usually not clear how big the perturbation step should be. We introduce a new property of fitness landscapes termed "Neighbours with Similar Fitness" and we demonstrate that the effectiveness of a restart strategy depends on this property.

[1]  Steven Okamoto,et al.  Explorative anytime local search for distributed constraint optimization , 2014, Artif. Intell..

[2]  Carsten Witt,et al.  Bioinspired Computation in Combinatorial Optimization , 2010, Bioinspired Computation in Combinatorial Optimization.

[3]  Bart Naudts,et al.  A comparison of predictive measures of problem difficulty in evolutionary algorithms , 2000, IEEE Trans. Evol. Comput..

[4]  Markus Wagner,et al.  An Improved Generic Bet-and-Run Strategy with Performance Prediction for Stochastic Local Search , 2019, AAAI.

[5]  Irene Moser,et al.  Identifying Features of Fitness Landscapes and Relating Them to Problem Difficulty , 2017, Evolutionary Computation.

[6]  Victor J. Rayward-Smith,et al.  Fitness Distance Correlation and Ridge Functions , 1998, PPSN.

[7]  Weinberger,et al.  Local properties of Kauffman's N-k model: A tunably rugged energy landscape. , 1991, Physical review. A, Atomic, molecular, and optical physics.

[8]  Rafael Martí.  Multi-Start Methods , 2003, Handbook of Metaheuristics.

[9]  Zbigniew J. Czech.  Statistical measures of a fitness landscape for the vehicle routing problem , 2008, 2008 IEEE International Symposium on Parallel and Distributed Processing.

[10]  Jinbo Huang,et al.  The Effect of Restarts on the Efficiency of Clause Learning , 2007, IJCAI.

[11]  Olivier Teytaud,et al.  A Rigorous Runtime Analysis for Quasi-Random Restarts and Decreasing Stepsize , 2011, Artificial Evolution.

[12]  Alistair Sinclair.  Markov chains and rapid mixing , 1993 .

[13]  Lee Altenberg,et al.  Fitness Distance Correlation Analysis: An Instructive Counterexample , 1997, ICGA.

[14]  Serdar Kadioglu,et al.  Parallel Restarted Search , 2014, AAAI.

[15]  E. Weinberger,et al.  Correlated and uncorrelated fitness landscapes and how to tell the difference , 1990, Biological Cybernetics.

[16]  Armin Biere.  Adaptive Restart Strategies for Conflict Driven SAT Solvers , 2008, SAT.

[17]  Markus Wagner,et al.  Heavy-Tailed Mutation Operators in Single-Objective Combinatorial Optimization , 2018, PPSN.

[18]  Bernd Freisleben,et al.  Fitness Landscapes, Memetic Algorithms, and Greedy Operators for Graph Bipartitioning , 2000, Evolutionary Computation.

[19]  Vassilis Zissimopoulos,et al.  Autocorrelation Coefficient for the Graph Bipartitioning Problem , 1998, Theor. Comput. Sci..

[20]  Carlos Ansótegui,et al.  Reactive Dialectic Search Portfolios for MaxSAT , 2017, AAAI.

[21]  Terry Jones,et al.  Fitness Distance Correlation as a Measure of Problem Difficulty for Genetic Algorithms , 1995, ICGA.

[22]  Dale Schuurmans,et al.  Local search characteristics of incomplete SAT procedures , 2000, Artif. Intell..

[23]  T. Stützle,et al.  Iterated Local Search: Framework and Applications , 2018, Handbook of Metaheuristics.

[24]  Martin C. Cooper,et al.  Tractable Constraints on Ordered Domains , 1995, Artif. Intell..

[25]  Thomas Jansen,et al.  On the Analysis of Dynamic Restart Strategies for Evolutionary Algorithms , 2002, PPSN.

[26]  Benjamin Doerr,et al.  Fast genetic algorithms , 2017, GECCO.

[27]  Peter Merz,et al.  Advanced Fitness Landscape Analysis and the Performance of Memetic Algorithms , 2004, Evolutionary Computation.

[28]  A. Pettitt,et al.  The Kolmogorov-Smirnov Goodness-of-Fit Statistic with Discrete and Grouped Data , 1977 .

[29]  Sébastien Vérel,et al.  Local Optima Networks of the Quadratic Assignment Problem , 2010, IEEE Congress on Evolutionary Computation.

[30]  Kishor S. Trivedi.  Probability and Statistics with Reliability, Queuing, and Computer Science Applications , 1984 .

[31]  Julian Francis Miller,et al.  Information Characteristics and the Structure of Landscapes , 2000, Evolutionary Computation.

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

[33]  P. Stadler.  Fitness Landscapes , 1993 .

[34]  Markus Wagner,et al.  Escaping large deceptive basins of attraction with heavy-tailed mutation operators , 2018, GECCO.

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

[36]  Irene Moser,et al.  Investigating the correlation between indicators of predictive diagnostic optimisation and search result quality , 2016, Inf. Sci..