Quantification-oriented learning based on reliable classifiers
作者：
Highlights：
• This paper studies the first quantification-oriented learning approach.
• We implement the first learning method that optimizes a quantification metric.
• We propose a new metric that combines quantification and classification.
• We compare our proposal with current quantifiers on benchmark datasets.
• Our method is theoretically well-founded and offers competitive performance.
摘要
Highlights•This paper studies the first quantification-oriented learning approach.•We implement the first learning method that optimizes a quantification metric.•We propose a new metric that combines quantification and classification.•We compare our proposal with current quantifiers on benchmark datasets.•Our method is theoretically well-founded and offers competitive performance.
论文关键词：Quantification,Class distribution estimation,Performance metrics,Reliability,Multivariate predictions
论文评审过程：Received 12 July 2013, Revised 13 June 2014, Accepted 31 July 2014, Available online 11 August 2014.
论文官网地址：https://doi.org/10.1016/j.patcog.2014.07.032
