A compressed sensing approach for efficient ensemble learning
作者：
Highlights：
• A compressed sensing approach for efficient ensemble learning is proposed.
• A new performance evaluation method (the roulette-wheel kappa-error) is proposed.
• The ensemble of classifiers is posed as a compressed sensing problem.
• Roulette-wheel is used to select pairs of classifiers based on their weightings.
• The proposed method is testified by 25 different public data sets.
摘要
Highlights•A compressed sensing approach for efficient ensemble learning is proposed.•A new performance evaluation method (the roulette-wheel kappa-error) is proposed.•The ensemble of classifiers is posed as a compressed sensing problem.•Roulette-wheel is used to select pairs of classifiers based on their weightings.•The proposed method is testified by 25 different public data sets.
论文关键词：Ensemble learning,Classification,Classifier ensemble,Sparse reconstruction,Compressed sensing,Roulette-wheel selection,Kappa-error
论文评审过程：Received 4 June 2013, Revised 5 March 2014, Accepted 16 April 2014, Available online 28 April 2014.
论文官网地址：https://doi.org/10.1016/j.patcog.2014.04.015
