Sebastian Herrmann
Franz Rothlauf
ABSTRACT
Previous studies have used statistical analysis of fitness landscapes such as ruggedness and deceptiveness in order to predict the expected quality of heuristic search methods. Novel approaches for predicting the performance of heuristic search are based on the analysis of local optima networks (LONs). A LON is a compressed stochastic model of a fitness landscape's basin transitions. Recent literature has suggested using various LON network measurements as predictors for local search performance.
In this study, we suggest PageRank centrality as a new measure for predicting the performance of heuristic search methods using local search. PageRank centrality is a variant of Eigenvector centrality and reflects the probability that a node in a network is visited by a random walk. Since the centrality of high-quality solutions in LONs determines the search difficulty of the underlying fitness landscape and since the big valley property suggests that local optima are not randomly distributed in the search space but rather clustered and close to one another, PageRank centrality can serve as a good predictor for local search performance. In our experiments for NK-models and the traveling salesman problem, we found that the PageRank centrality is a very good predictor for the performance of first-improvement local search as well as simulated annealing, since it explains more than 90% of the variance of search performance. Furthermore, we found that PageRank centrality is a better predictor of search performance than traditional approaches such as ruggedness, deceptiveness, and the length of the shortest path to the optimum.
- K. D. Boese, A. B. Kahng, and S. Muddu. A new adaptive multi-start technique for combinatorial global optimizations. Operations Research Letters, 16(2):101--113, 1994. Google Scholar
Digital Library
- S. P. Borgatti. Centrality and network flow. Social Networks, 27(1):55--71, Jan. 2005.Google ScholarCross Ref
- S. Brin and L. Page. The anatomy of a large-scale hypertextual Web search engine. Computer Networks and ISDN Systems, 30(1--7):107--117, Apr. 1998. Google ScholarDigital Library
- S. Choi and B. Moon. Normalization for Genetic Algorithms With Nonsynonymously Redundant Encodings. IEEE Transactions on Evolutionary Computation, 12(5):604--616, 2008. Google ScholarDigital Library
- F. Daolio, S. Verel, G. Ochoa, and M. Tomassini. Local optima networks and the performance of iterated local search. Proceedings of the fourteenth international conference on Genetic and evolutionary computation conference - GECCO '12, page 369, Oct. 2012. Google ScholarDigital Library
- M. Franceschet. PageRank: standing on the shoulders of giants. Communications of the ACM, 54(6):92, June 2011. Google ScholarDigital Library
- L. C. Freeman. The Development of Social Network Analysis. Createspace, 2004.Google Scholar
- G. Frobenius. Über Matrizen aus nicht negativen Elementen. Sitzungsberichte Preussische Akademie der Wissenschaft, Berlin, pages 456--477, 1912.Google Scholar
- F. Glover. Future paths for integer programming and links to artificial intelligence. Computers & Operations Research, 13(5):533--549, Jan. 1986. Google ScholarDigital Library
- J. He, C. Reeves, C. Witt, and X. Yao. A note on problem difficulty measures in black-box optimization: classification, realizations and predictability. IEEE Transactions on Evolutionary Computation, 15(4):435--43, Jan. 2007. Google ScholarDigital Library
- T. Jones and S. Forrest. Fitness Distance Correlation as a Measure of Problem Difficulty for Genetic Algorithms. Proceedings of the Sixth International Conference on Genetic Algorithms, 1995. Google ScholarDigital Library
- L. Kallel, B. Naudts, and C. R. Reeves. Properties of Fitness Functions and Search Landscapes. In L. Kallel, B. Naudts, and A. Rogers, editors, Theoretical Aspects of Evolutionary Computing, Natural Computing Series, pages 175--206. Springer Berlin Heidelberg, 2001. Google ScholarDigital Library
- S. Kauffman and S. Levin. Towards a General Theory of Adaptive Walks on Rugged Landscapes. Journal of theoretical Biology, pages 11--45, 1987.Google Scholar
- S. Kirkpatrick, C. D. Gelatt, and M. P. Vecchi. Optimization by simulated annealing. Science (N.Y.), 220(4598):671--80, May 1983.Google Scholar
- P. J. M. Laarhoven and E. H. L. Aarts. Simulated Annealing: Theory and Applications. Kluwer Academic Publishers, Norwell, MA, USA, 1988. Google ScholarDigital Library
- D. H. Lehmer. Teaching combinatorial tricks to a computer. In Proceedings of the Symposium on Applied Mathematics and Combinatorial Analysis., 1960.Google ScholarCross Ref
- J. K. Lenstra and a.H. Rinnooy Kan. Complexity of Vehicle Routing and Scheduling Problems. Networks, 11:221--227, 1981.Google ScholarCross Ref
- K. M. Malan and A. P. Engelbrecht. Fitness Landscape Analysis for Metaheuristic Performance Prediction. In H. Richter and A. Engelbrecht, editors, Recent Advances in the Theory and Application of Fitness Landscapes, pages 103--129. Springer, 2014.Google ScholarCross Ref
- G. Ochoa, M. Tomassini, S. Vérel, and C. Darabos. A study of NK landscapes' basins and local optima networks. In Proceedings of the 10th annual conference on Genetic and evolutionary computation - GECCO '08, page 555, New York, 2008. ACM Press. Google ScholarDigital Library
- G. Ochoa, S. Verel, and M. Tomassini. First-improvement vs. best-improvement local optima networks of nk landscapes. PPSN'10 Proceedings of the 11th international conference on Parallel problem solving from nature, I:104--113, 2010. Google ScholarDigital Library
- O. Perron. Zur Theorie der Matrices. Mathematische Annalen1, 64(1):248--263, 1907.Google ScholarCross Ref
- E. Pitzer and M. Affenzeller. A Comprehensive Survey on Fitness Landscape Analysis. In J. Fodor, R. Klempous, and C. P. Suárez Araujo, editors, Recent Advances in Intelligent Engineering Systems, volume 378 of Studies in Computational Intelligence, pages 161--191. Springer, Berlin and Heidelberg, 2012.Google ScholarCross Ref
- C. M. Reidys and P. F. Stadler. Combinatorial Landscapes. SIAM Review, 44(1):3--54, Jan. 2002. Google ScholarDigital Library
- F. Rothlauf. Design of modern heuristics: Principles and application. Springer, Berlin and Heidelberg, 2011. Google ScholarCross Ref
- P. F. Stadler. Landscapes and their correlation functions. Journal of Mathematical Chemistry, 20(1):1--45, 1996.Google ScholarCross Ref
- F. H. Stillinger. A topographic view of supercooled liquids and glass formation. Science (N.Y.), 267(5206):1935--9, Mar. 1995.Google Scholar
- E. Weinberger. Correlated and Uncorrelated Fitness Landscapes and How to Tell the Difference. Biological cybernetics, 336:325--336, 1990. Google ScholarDigital Library
- S. Wright. The roles of mutation, inbreeding, crossbreeding, and selection in evolution. In Proceedings of the 6th International Congress of Genetics, pages 356--366, 1932.Google Scholar
Index Terms
- Predicting Heuristic Search Performance with PageRank Centrality in Local Optima Networks
Recommendations
Local optima networks for continuous fitness landscapes
GECCO '19: Proceedings of the Genetic and Evolutionary Computation Conference CompanionLocal Optima Networks (LONs) have been proposed as a coarsegrained model of discrete (combinatorial) fitness landscapes, where nodes are local optima and edges are search transitions based on an exploration search operator. This paper presents one of ...
Communities of Local Optima as Funnels in Fitness Landscapes
GECCO '16: Proceedings of the Genetic and Evolutionary Computation Conference 2016We conduct an analysis of local optima networks extracted from fitness landscapes of the Kauffman NK model under iterated local search. Applying the Markov Cluster Algorithm for community detection to the local optima networks, we find that the ...
PageRank centrality for performance prediction: the impact of the local optima network model
A local optima network (LON) compresses relevant features of fitness landscapes in a complex network, where nodes are local optima and edges represent transition probabilities between different basins of attraction. Previous work has found that the ...
Comments