STATISTICAL LEARNING THEORY

 

Currently I am working in Statistical learning theory.  The learning problem from examples is simple to formulate. Suppose, we have a finite number of (x,y) pairs, where the y-values are supposed to have been obtained from the respective x-values by applying a function f. Since the function f is unknown, the problem is to find an approximation of this unknown function f. The approximation can be done in many ways imposing familiar concepts of closeness. Further restrictions of approximating f from a given function space or to satisfy some other statistical criteria such as predictivity etc. makes the mathematics of learning.




PUBLICATIONS

  • Sahoo, J. K., Singh, A., An error analysis of Lavrentiev regularization in learning theory,

    Asian-European Journal of Mathematics, 2(1):129-140, 2009.

  • Sahoo, J. K., Singh, A., An estimate of misclassification error with Hinge and square loss, Int. J. Comput. Appl. Math. 4(2):173-186, 2009.

  • Sahoo, J. K., Singh, A., An improved error estimate for classification problem, (Communicated).

  • Sahoo, J. K., Singh, A., Reconstruction of function and regression estimation from the sparse data, Proceedings of International  conference on new trends in statistics and optimization,Srinagar, India. (To appear).