A unified pipeline for online feature selection and classification
作者：
Highlights：
• A proposal for online feature selection is proposed.
• The proposed pipeline covers discretization, feature selection and classification.
• Classical algorithms were modified to make them work online.
• K-means discretizer, Chi-Square filter and Artificial Neural Networks were used.
• Results show that classification error is decreasing, adapting to the arrival of new data.
摘要
•A proposal for online feature selection is proposed.•The proposed pipeline covers discretization, feature selection and classification.•Classical algorithms were modified to make them work online.•K-means discretizer, Chi-Square filter and Artificial Neural Networks were used.•Results show that classification error is decreasing, adapting to the arrival of new data.
论文关键词：Machine learning,Online learning,Classification,Feature selection,Discretization
论文评审过程：Received 16 December 2014, Revised 27 January 2016, Accepted 15 February 2016, Available online 27 February 2016, Version of Record 15 March 2016.
论文官网地址：https://doi.org/10.1016/j.eswa.2016.02.035
