Teaching Learning Based Optimized Mathematical Model for Data Classification Problems
This paper presents application of yet one more optimization technique based on evolutionary computation approach to locate the optimal values of the coefficients of terms of polynomial equations which is developed to classify the unknown dataset. A recent optimization technique known as Teaching Learning Based Optimization (TLBO) is used here for optimizing the coefficients of polynomial terms for classifying many bench mark datasets. The original mathematical model for classification problems are developed using Polynomial neural network (PNN) and then the coefficients of the terms of polynomials are optimized separately with Least mean square (LSE) and TLBO approach. The comparisons of the two approaches are suitably presented with classification accuracies for many bench mark datasets. The results reveal that TLBO optimized polynomials are performing better than LSE- optimized polynomial, herein known as PNN models for all investigated datasets.
Unable to display preview. Download preview PDF.
- Breitman, L., Friedman, J.H., Olshen, R.A., Stone, C.J.: Classification and Regression tress. Wadsworth, Belmont (1984)Google Scholar
- Hanson, R., Stutz, J., Cheeseman, P.: Bayesian classification with correlation and inheritance. In: Proceedings of the 12th International Joint Conference on Artificial Intelligence, Sydney, Australia, vol. 2, pp. 692–698. Morgan Kaufmann (1992)Google Scholar
- Michie, D., et al.: Machine Learning, Neural and Statistical Classification, Ellis Horwood (1994)Google Scholar
- Tsoi, A.C., et al.: Comparison of three classification Techniques, CART, C4.5 and multilayer perceptrons. In: Advances in Neural Information Processing Systems, vol. 3, pp. 963–969 (1991)Google Scholar
- Adem, J., Gochet, W.: Mathematical Program based heuristics for improving LP-generated classifiers for the multi-class supervised classification problem. European Journal of Operation Research (1994)Google Scholar
- Misra, B.B., Satapathy, S.C., Biswal, B.N., Dash, P.K., Panda, G.: Pattern Classification using Polynomial Neural Network. Accepted for presentation at IEEE International Conferences on Cybernetics & Intelligent Systems (CIS) and Robotics, Automation & Mechatronics (RAM), CIS-RAM 2006 (2006)Google Scholar