Centralized vs. distributed feature selection methods based on data complexity measures

详细信息    查看全文

• 作者：L. Mor&aacute ; n-Fern&aacute ; ndez ; ^{laura.moranf@udc.es} ; V. Boló ; n-Canedo ^{vbolon@udc.es} ; A. Alonso-Betanzos ^{ciamparo@udc.es}
• 关键词：Distributed learning ; Feature selection ; Data complexity measures ; Classification
• 刊名：Knowledge-Based Systems
• 出版年：2017
• 出版时间：1 February 2017
• 年：2017
• 卷：117
• 期：Complete
• 页码：27-45
• 全文大小：1072 K
• 卷排序：117

文摘

A methodology for distributing the process of feature selection based on several data complexity measures is proposed. We tackled the two strategies to partition the datasets: horizontal (i.e. by samples) and vertical (i.e. by features). We present an experimental study on 11 datasets (five of them microarrays) in terms of number of selected features, classification accuracy and running time. The novel procedures are able to reduce significantly the running time while maintaining (or even improving) the classification performance.