Performance improvements of a Kohonen self organizing classification algorithm on sparse data sets

ABSTRACT

This paper presents a variation of a Kohonen self organizing feature map. From the proposed algorithm possible performance improvements are investigated in terms of time and space complexity taking advantage from a sparse input data set. The proposed variation has been tested on different datasets coming from case studies in the field of bioinformatics. The improvements make the application of the algorithms feasible to massive document collections. The application of the proposed improvements for grid implementations could be beneficial to reduce the computing element demand.

References

1. Y. Zhao, G. Karypis, "Data clustering in life science", Molecular Biotechnology, vol. 31, no. 1, 2005, pp. 55-80.
2. R. Xu, D. Wunsch "Survey of Clustering Algorithms", IEEE Transactions on Neural Networks, vol. 16, no. 3, 2005.
3. T. Kohonen, "Self Organizing Maps", Springer 1995.
4. S. Kaski, J. Kangas, T. Kohonen, "Bibliography of self organizing map (SOM) papers: 1981-1997, Neural Computing Survey, vol. 1, no. 3, 1998, pp. 102-350.
5. M. Oja, S. Kaski, T. Kohonen, "Bibliography of self organizing map (SOM) papers: 1998-2001 Addendum, Neural Computing Survey, vol. 3, no. 1, 2003, pp. 1-156.
6. M. Cottrel, J. C. Fort, P. Letremy, "Advantages and drawbacks of the batch Kohonen Algorithm", 10th European Symp. On Artificial Neural Network. Bruges (Belgium), 2005, pp. 223-230.
7. A. Faro, D. Giordano, F. Maiorana, "Discovering complex regularities by adaptive Self Organizing classification", Enformatika, vol. I, 2005, pp. 27--30.
8. A. Faro, D. Giordano, F. Maiorana, "Discovering complex regularities from tree to semi - lattice classifications". International Journal of Computational Intelligence, vol. 2, no. 1, 2005, pp. 34-39.
9. E. C. Vargas, R. Francelin Romero, K. Obermayer, "Speeding up algorithms for SOM family for large and high dimensional databases", Proceedings of the Workshop on Self Organizing Maps Hibikino (Japan), 2003, pp. 167-172.
10. R. D. Lawrence, G. S. Almasi, H. F. Rushmejer, "A scalable parallel algorithm for Self organizing maps with applications to sparse data mining problems", Data Mining and Knowledge Discovery, vol. 3, no. 171, 1999, pp 171-195.
11. R. Natarajan, "Exploratory data analysis in large sparse datasets", IBM Research Report RC 20749, IBM Research, Yorktown Heights, NY, 1997.
12. Z. Zhao, "Improvements to Kohonen self-organising algorithm", Electronics Letters, vol. 30, no. 6, 1994, pp. 502-503.
13. T. Kohonen, T. "Speedup of SOM Computation".
14. B. K. Y. Chan, W. W. S. Chu, L. Xu, "Empirical comparison between two computational strategies for topological self-organization", Intelligent Data Engineering and Automated Learning (LNCS), vol 2690, Springer, 2003, pp. 410-414.
15. B. C. Guez, F. Rossi, A. E. Golli, "Fast algorithm and implementation of dissimilarity self organizing maps", Neural Networks, vol. 19, 2006, pp. 855-863.
16. B. C. Guez, F. Rossi, "Speeding up the dissimilarity Self Organizing Maps by Branch and Bound", Computational and Ambient Intelligence (LNCS), vol. 4507, Springer, 2007, pp 203-210.
17. C. Wei, Y. Lee, C. Hsu, "Empirical comparison of fast partitioning-based clustering algorithms for large data sets", Expert Systems with applications, vol 24, 2003, pp. 351-363.
18. A. El Golli, "Speeding up the self organizing map for dissimilarity data", Proceedings of International Symposium on Applied Stochastic Models and Data Analysis, Brest, France, 2005, pp. 709-713.
19. M. Nocker, F. Morchen, A. Ultsch, "An algorithm for fast and reliable ESOM learning", Proceedings of 14th European Symposium on Artificial Neural Networks, Bruges, Belgium, 2006, pp. 131-136.
20. A. Faro, D. Giordano, F. Maiorana, C. Spampinato, "Discovering Genes-Diseases Associations from Specialized Literature using the GRID", to appear on IEEE Transaction on Information Technology in Biomedicine.

Index Terms

• Computing methodologies

  • Machine learning

    • Learning paradigms

      • Supervised learning

        • Supervised learning by classification

      • Unsupervised learning

        • Cluster analysis

    • Machine learning approaches

      • Classification and regression trees

      • Neural networks

  • Modeling and simulation

    • Model development and analysis

      • Model verification and validation

• Hardware

  • Emerging technologies

    • Circuit substrates

      • Cellular neural networks

• Theory of computation

  • Theory and algorithms for application domains

    • Machine learning theory