Quadratically constrained maximum a posteriori estimation for binary classifier

Authors:
Tatsuya Yokota;Yukihiko Yamashita
Affiliations:
Tokyo Institute of Technology, Tokyo, Japan;Tokyo Institute of Technology, Tokyo, Japan
Venue:
MLDM'11 Proceedings of the 7th international conference on Machine learning and data mining in pattern recognition
Year:
2011

Citing 8
Cited 0

Feasibility issues in a primal-dual interior-point method for linear programming

Mathematical Programming: Series A and B
A maximum a posteriori estimate for the source separation problem with statistical knowledge about the mixing matrix

Pattern Recognition Letters
Preventing Over-Fitting during Model Selection via Bayesian Regularisation of the Hyper-Parameters

The Journal of Machine Learning Research
A novel kernel-based maximum a posteriori classification method

Neural Networks
A maximum a-posteriori identification criterion for wavelet domain watermarking

International Journal of Wireless and Mobile Computing
Maximum a posteriori based kernel classifier trained by linear programming

SSPR&SPR'10 Proceedings of the 2010 joint IAPR international conference on Structural, syntactic, and statistical pattern recognition
Maximum a posterior probability and cumulative distribution function equalization methods for speech spectral estimation with application in noise suppression filtering

NOLISP'05 Proceedings of the 3rd international conference on Non-Linear Analyses and Algorithms for Speech Processing
Cost functions to estimate a posteriori probabilities in multiclass problems

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we propose a new classification criterion based on maximum a posteriori (MAP) estimation for a binary problem. In our method, we do not estimate the posteriori probability; instead we construct a discriminant function that provides the same result. The criterion consists of the maximization of an expected cost function and a quadratic constraint of the discriminant function with a weighting function. By selecting different weighting functions we show that the least squares regression and the support vector machine can be derived from the criterion. Furthermore, we propose a novel classifier based on the criterion and conduct experiments to demonstrate its advantages.