Feature selection for semi-supervised multi-label learning with application to gene function analysis

Authors:
Guo-Zheng Li;Mingyu You;Lei Ge;Jack Y. Yang;Mary Qu Yang
Affiliations:
Tongji University, Shanghai, China;Tongji University, Shanghai, China;Shanghai University, Shanghai, China;Indiana Univ Bioinformatics Center, Indianapolis, IN;Indiana Univ Bioinformatics Center, Indianapolis, IN
Venue:
Proceedings of the First ACM International Conference on Bioinformatics and Computational Biology
Year:
2010

Citing 7
Cited 0

Combining labeled and unlabeled data with co-training

COLT' 98 Proceedings of the eleventh annual conference on Computational learning theory
Semi-supervised learning with graphs

Semi-supervised learning with graphs
Hierarchical multi-label prediction of gene function

Bioinformatics
ML-KNN: A lazy learning approach to multi-label learning

Pattern Recognition
Document Transformation for Multi-label Feature Selection in Text Categorization

ICDM '07 Proceedings of the 2007 Seventh IEEE International Conference on Data Mining
Semi-supervised regression with co-training

IJCAI'05 Proceedings of the 19th international joint conference on Artificial intelligence
Feature selection for bagging of support vector machines

PRICAI'06 Proceedings of the 9th Pacific Rim international conference on Artificial intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper investigates gene function annotation of Yeast by using semi-supervised multi-label learning. Multi-label learning has been a hot topic in the bioinformatics field, but there are many samples unlabeled. Semi-supervised learning may be employed to utilize the unlabeled data. This paper proposes a novel semi-supervised multi-label learning algorithm COMN by combining Co-Training with ML-kNN to utilize the unlabeled yeast gene data to improve modeling accuracy of function annotation. Furthermore, an embedded feature selection algorithm PRECOMN is proposed to perform feature selection for COMN to remove the irrelevant and redundant features. Experimental results on one benchmark data set of Yeast show COMN and PRECOMN perform better than the original multi-label learning algorithm ML-kNN. Furthermore PRECOMN improves generalization performance of COMN.