Multi-task optimisation for multi-objective feature selection in classification

Jiabin Lin, Qi Chen, Bing Xue, Mengjie Zhang

Abstract: Many effective evolutionary multi-objective feature selection algorithms have been developed in recent years. However, most of them tend to address feature selection tasks independently, while in real-world applications, many feature selection tasks are closely related to each other and share common knowledge. Multi-task optimisation, which aims to address multiple related optimisation tasks simultaneously and share common knowledge across them, can benefit feature selection. However, it is seldom considered for feature selection. In this work, we develop a multi-task multi-objective optimisation algorithm for feature selection in classification, with the aim of capturing and sharing common knowledge for related feature selection tasks. To evaluate the effectiveness of the proposed algorithm, we conduct a set of experiments to compare its performance with that of the single-task multi-objective feature selection algorithm on three sets of related feature selection tasks. With the help of knowledge transfer, our new algorithm significantly improved the feature selection performance is more efficient.