Reducing bias in supervised learning

Nonparametric statistical supervised learning methods often suffer from bias caused by non-uniformity of the probability distribution of training samples. This problem is discussed in this paper and a new nonparametric neighborhood method for classification and estimation that significantly reduces the bias is proposed. Simulations exemplify the advantages, and theoretical results are noted.

[1]  Roberto Guerrieri,et al.  A Geometric Approach to Maximum-Speed n-Dimensional Continuous Linear Interpolation in Rectangular Grids , 1998, IEEE Trans. Computers.

[2]  Jerome H. Friedman,et al.  On Bias, Variance, 0/1—Loss, and the Curse-of-Dimensionality , 2004, Data Mining and Knowledge Discovery.

[3]  E. Jaynes.  Information Theory and Statistical Mechanics , 1957 .

[4]  Nicholas Kalouptsidis,et al.  Nearest neighbor pattern classification neural networks , 1994, Proceedings of 1994 IEEE International Conference on Neural Networks (ICNN'94).

[5]  T. Kohonen,et al.  Statistical pattern recognition with neural networks: benchmarking studies , 1988, IEEE 1988 International Conference on Neural Networks.

[6]  Maya R. Gupta,et al.  Analysis and classification of internal pipeline images , 2003, Proceedings 2003 International Conference on Image Processing (Cat. No.03CH37429).