Understanding TSP Difficulty by Learning from Evolved Instances

  • Kate Smith-Miles
  • Jano van Hemert
  • Xin Yu Lim
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6073)


Whether the goal is performance prediction, or insights into the relationships between algorithm performance and instance characteristics, a comprehensive set of meta-data from which relationships can be learned is needed. This paper provides a methodology to determine if the meta-data is sufficient, and demonstrates the critical role played by instance generation methods. Instances of the Travelling Salesman Problem (TSP) are evolved using an evolutionary algorithm to produce distinct classes of instances that are intentionally easy or hard for certain algorithms. A comprehensive set of features is used to characterise instances of the TSP, and the impact of these features on difficulty for each algorithm is analysed. Finally, performance predictions are achieved with high accuracy on unseen instances for predicting search effort as well as identifying the algorithm likely to perform best.


Algorithm Selection Travelling Salesman Problem Hardness Prediction Phase Transition Combinatorial optimisation Instance Difficulty 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Macready, W., Wolpert, D.: What makes an optimization problem hard. Complexity 5, 40–46 (1996)MathSciNetGoogle Scholar
  2. 2.
    Nudelman, E., Leyton-Brown, K., Hoos, H., Devkar, A., Shoham, Y.: Understanding random SAT: Beyond the clauses-to-variables ratio. In: Wallace, M. (ed.) CP 2004. LNCS, vol. 3258, pp. 438–452. Springer, Heidelberg (2004)Google Scholar
  3. 3.
    Xu, L., Hutter, F., Hoos, H., Leyton-Brown, K.: SATzilla-07: The design and analysis of an algorithm portfolio for SAT. In: Bessière, C. (ed.) CP 2007. LNCS, vol. 4741, pp. 712–727. Springer, Heidelberg (2007)CrossRefGoogle Scholar
  4. 4.
    Cho, Y., Moore, J., Hill, R., Reilly, C.: Exploiting empirical knowledge for bi-dimensional knapsack problem heuristics. International Journal of Industrial and Systems Engineering 3(5), 530–548 (2008)CrossRefGoogle Scholar
  5. 5.
    Hall, N., Posner, M.: Performance Prediction and Preselection for Optimization and Heuristic Solution Procedures. Operations Research 55(4), 703 (2007)zbMATHCrossRefGoogle Scholar
  6. 6.
    Smith-Miles, K.: Towards insightful algorithm selection for optimisation using meta-learning concepts. In: IEEE International Joint Conference on Neural Networks, IJCNN 2008. IEEE World Congress on Computational Intelligence, pp. 4118–4124 (2008)Google Scholar
  7. 7.
    Smith-Miles, K., James, R., Giffin, J., Tu, Y.: Understanding the Relationship between Scheduling Problem Structure and Heuristic Performance using Knowledge Discovery, LNCS. Springer, Heidelberg (in press, 2009)Google Scholar
  8. 8.
    Rice, J.: The Algorithm Selection Problem. Advances in computers 65 (1976)Google Scholar
  9. 9.
    van Hemert, J.: Evolving combinatorial problem instances that are difficult to solve. Evolutionary Computation 14(4), 433–462 (2006)CrossRefGoogle Scholar
  10. 10.
    Gras, R.: How efficient are genetic algorithms to solve high epistasis deceptive problems? In: IEEE Congress on Evolutionary Computation, CEC 2008. IEEE World Congress on Computational Intelligence, pp. 242–249 (2008)Google Scholar
  11. 11.
    Locatelli, M., Wood, G.: Objective Function Features Providing Barriers to Rapid Global Optimization. Journal of Global Optimization 31(4), 549–565 (2005)zbMATHCrossRefMathSciNetGoogle Scholar
  12. 12.
    Xin, B., Chen, J., Pan, F.: Problem difficulty analysis for particle swarm optimization: deception and modality. In: Proceedings of the first ACM/SIGEVO Summit on Genetic and Evolutionary Computation, pp. 623–630 (2009)Google Scholar
  13. 13.
    Bachelet, V.: Métaheuristiques parallèles hybrides: application au problème d’affectation quadratique. PhD thesis, Universite des Sciences et Technologies de Lille (1999)Google Scholar
  14. 14.
    Reeves, C.: Landscapes, operators and heuristic search. Annals of Operations Research 86, 473–490 (1999)zbMATHCrossRefMathSciNetGoogle Scholar
  15. 15.
    Schiavinotto, T., Stützle, T.: A review of metrics on permutations for search landscape analysis. Comput. Oper. Res. 34(10), 3143–3153 (2007)zbMATHCrossRefGoogle Scholar
  16. 16.
    Smith-Miles, K.A., Lopes, L.B.: Measuring Combinatorial Optimization Problem Difficulty for Algorithm Selection. Annals of Mathematics and Artificial Intelligence (under review, 2009)Google Scholar
  17. 17.
    Pfahringer, B., Bensusan, H., Giraud-Carrier, C.: Meta-learning by landmarking various learning algorithms. In: Proceedings of the Seventeenth International Conference on Machine Learning table of contents, pp. 743–750. Morgan Kaufmann Publishers Inc., San Francisco (2000)Google Scholar
  18. 18.
    Burke, E., Kendall, G., Newall, J., Hart, E., Ross, P., Schulenburg, S.: Hyper-heuristics: An emerging direction in modern search technology. International Series in Operations Research and Management Science, pp. 457–474 (2003)Google Scholar
  19. 19.
    Battiti, R.: Using mutual information for selecting features in supervised neural net learning. IEEE Transactions on neural networks 5(4), 537–550 (1994)CrossRefGoogle Scholar
  20. 20.
    Vasconcelos, N.: Feature selection by maximum marginal diversity: optimality and implications for visual recognition. In: Proceedings of 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 1 (2003)Google Scholar
  21. 21.
    Cheeseman, P., Kanefsky, B., Taylor, W.: Where the really hard problems are. In: Proceedings of the 12th IJCAI, pp. 331–337 (1991)Google Scholar
  22. 22.
    Ridge, E., Kudenko, D.: An Analysis of Problem Difficulty for a Class of Optimisation Heuristics. In: Cotta, C., van Hemert, J. (eds.) EvoCOP 2007. LNCS, vol. 4446, pp. 198–209. Springer, Heidelberg (2007)CrossRefGoogle Scholar
  23. 23.
    Zhang, W., Korf, R.: A study of complexity transitions on the asymmetric traveling salesman problem. Artificial Intelligence 81(1-2), 223–239 (1996)CrossRefMathSciNetGoogle Scholar
  24. 24.
    Zhang, W.: Phase transitions and backbones of the asymmetric traveling salesman problem. Journal of Artificial Intelligence Research 21, 471–497 (2004)zbMATHMathSciNetGoogle Scholar
  25. 25.
    Gent, I., Walsh, T.: The TSP phase transition. Artificial Intelligence 88(1-2), 349–358 (1996)zbMATHCrossRefMathSciNetGoogle Scholar
  26. 26.
    Thiebaux, S., Slaney, J., Kilby, P.: Estimating the hardness of optimisation. In: ECAI, pp. 123–130 (2000)Google Scholar
  27. 27.
    Gaertner, D., Clark, K.: On optimal parameters for ant colony optimization algorithms. In: Proceedings of the 2005 International Conference on Artificial Intelligence, Citeseer, vol. 1, pp. 83–89 (2005)Google Scholar
  28. 28.
    Stadler, P., Schnabl, W.: The landscape of the traveling salesman problem. Phys. Lett. A 161(4), 337–344 (1992)zbMATHCrossRefMathSciNetGoogle Scholar
  29. 29.
    Kilby, P., Slaney, J., Walsh, T.: The backbone of the travelling salesperson. In: International Joint Conference on Artificial Intelligence, vol. 19, p. 175 (2005)Google Scholar
  30. 30.
    Lin, S., Kernighan, B.: An efficient heuristic algorithm for the traveling salesman problem. Operations Research 21(2) (1973)Google Scholar
  31. 31.
    van Hemert, J.: Property analysis of symmetric travelling salesman problem instances acquired through evolution. In: Raidl, G.R., Gottlieb, J. (eds.) EvoCOP 2005. LNCS, vol. 3448, pp. 122–131. Springer, Heidelberg (2005)CrossRefGoogle Scholar
  32. 32.
    van Hemert, J., Urquhart, N.: Phase transition properties of clustered travelling salesman problem instances generated with evolutionary computation. In: Yao, X., Burke, E.K., Lozano, J.A., Smith, J., Merelo-Guervós, J.J., Bullinaria, J.A., Rowe, J.E., Tiňo, P., Kabán, A., Schwefel, H.-P. (eds.) PPSN 2004. LNCS, vol. 3242, pp. 151–160. Springer, Heidelberg (2004)Google Scholar
  33. 33.
    van Hemert, J.: Property analysis of symmetric travelling salesman problem instances acquired through evolution. In: Raidl, G.R., Gottlieb, J. (eds.) EvoCOP 2005. LNCS, vol. 3448, pp. 122–131. Springer, Heidelberg (2005)CrossRefGoogle Scholar
  34. 34.
    Kratica, J., Ljubić, I., Tošic, D.: A genetic algorithm for the index selection problem. In: Raidl, G.R., Cagnoni, S., Cardalda, J.J.R., Corne, D.W., Gottlieb, J., Guillot, A., Hart, E., Johnson, C.G., Marchiori, E., Meyer, J.-A., Middendorf, M. (eds.) EvoIASP 2003, EvoWorkshops 2003, EvoSTIM 2003, EvoROB/EvoRobot 2003, EvoCOP 2003, EvoBIO 2003, and EvoMUSART 2003. LNCS, vol. 2611, pp. 281–291. Springer, Heidelberg (2003)CrossRefGoogle Scholar
  35. 35.
    Lin, S., Kernighan, B.: An effective heuristic algorithm for the traveling salesman problem. Operations Research 21, 498–516 (1973)zbMATHCrossRefMathSciNetGoogle Scholar
  36. 36.
    Applegate, D., Cook, W., Rohe, A.: Chained lin-kernighan for large travelling salesman problems (2000),
  37. 37.
    Johnson, D., McGeoch, L.: The traveling salesman problem: a case study. In: Aarts, E., Lenstra, J. (eds.) Local Search in Combinatorial Optimization, pp. 215–310. John Wiley & Sons, Inc., Chichester (1997)Google Scholar
  38. 38.
    Sander, J., Ester, M., Kriegel, H., Xu, X.: Density-based clustering in spatial databases: The algorithm gdbscan and its applications. Data Mining and Knowledge Discovery 2(2), 169–194 (1998)CrossRefGoogle Scholar
  39. 39.
    Kohonen, T.: Self-organization maps. Proc. IEEE 78, 1464–1480 (1990)CrossRefGoogle Scholar
  40. 40.
    SOMine, V.: Enterprise Edition Version 3.0, Eudaptics Software Gmbh (1999)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2010

Authors and Affiliations

  • Kate Smith-Miles
    • 1
  • Jano van Hemert
    • 2
  • Xin Yu Lim
    • 1
    • 3
  1. 1.School of Mathematical SciencesMonash UniversityVictoriaAustralia
  2. 2.School of InformaticsUniversity of EdinburghEdinburghUK
  3. 3.Mathematical InstituteOxford UniversityOxfordUK

Personalised recommendations