A Label Extended Semi-supervised Learning Method for Drug-target Interaction Prediction

DOI

10.2991/amcce-15.2015.292How to use a DOI?

Keywords

Data mining; Drug-target prediction; Label extension; Semi-supervised learning

Abstract

Computational methods for predicting the new drug-target interactions are more efficient that those experimental methods. Many machine learning based methods have been proposed but most of them suffer from false negative problem. In this paper we extend the original label matrix and adopt weighted lose function to overcome the traditional false negative problem and then propose a label extended semi-supervised learning method called LESSL for drug-target prediction. In our experiment we use two kinds of cross-validation.The results show that our method can raise AUC average by 0.03 and raise AUPR average by 0.04. At last we use the whole dataset as a training set and predict over 10 new drug-target interactions.To conclude our method is efficient and practicable.

Copyright

© 2015, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Jie Zhao
AU  - Zhi Cao
PY  - 2015/04
DA  - 2015/04
TI  - A Label Extended Semi-supervised Learning Method for Drug-target Interaction Prediction
BT  - Proceedings of the 2015 International Conference on Automation, Mechanical Control and Computational Engineering
PB  - Atlantis Press
SN  - 1951-6851
UR  - https://doi.org/10.2991/amcce-15.2015.292
DO  - 10.2991/amcce-15.2015.292
ID  - Zhao2015/04
ER  -