Classification of bio-data with small data set using additive factor model and SVM

Authors:
Hyeyoung Park;Minkook Cho
Affiliations:
School of Electrical Engineering and Computer Science, Kyungpook National University, Deagu, Korea;School of Electrical Engineering and Computer Science, Kyungpook National University, Deagu, Korea
Venue:
AI'06 Proceedings of the 19th Australian joint conference on Artificial Intelligence: advances in Artificial Intelligence
Year:
2006

Citing 7
Cited 2

Introduction to statistical pattern recognition (2nd ed.)

Introduction to statistical pattern recognition (2nd ed.)
An information-maximization approach to blind separation and blind deconvolution

Neural Computation
Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection

IEEE Transactions on Pattern Analysis and Machine Intelligence
Neural Networks for Pattern Recognition

Neural Networks for Pattern Recognition
High Confidence Visual Recognition of Persons by a Test of Statistical Independence

IEEE Transactions on Pattern Analysis and Machine Intelligence
Separating Style and Content with Bilinear Models

Neural Computation
A robust SVM design for multi-class classification

AI'05 Proceedings of the 18th Australian Joint conference on Advances in Artificial Intelligence

A Subspace Method Based on Data Generation Model with Class Information

Neural Information Processing
A feature analysis for dimension reduction based on a data generation model with class factors and environment factors

Computer Vision and Image Understanding

Quantified Score

Hi-index	0.00

Visualization

Abstract

Bio-data, which are obtained from human individuals, have been one of main applications of pattern classification these days. A critical property of bio-data classification is the small number of data in each class due to high cost of obtaining data from each individuals. Since most classification methods are based on the distribution of data in each class, the lack of data can be a main cause of low classification performance of conventional classifiers. To solve this problem, we propose a modified additive factor model for bio-data which has two factors; the individual factor and the environment factor. Under the proposed model, we estimate the distribution of environment factor which gives robust information even in case of small data set. We then define new similarity measures using the information. The similarity measure is applied to nearest neighbor method for classification. We also use the support vector machines (SVM) to find a sophisticated similarity measure. Through computational experiments, we confirm that the proposed model and similarity measure is appropriate enough to show better classification performance compared to conventional similarity measure as well as conventional SVM classifier.