Growing construction of conlitron and multiconlitron

作者：

Highlights：

• Multiconlitron is a general piecewise linear classifier as a union of conlitrons.

• A new technique is proposed for improving the performance of a conlitron/multiconlitron.

• This technique consists of two basic operations, SQUEEZE and INFLATE.

• SQUEEZE and INFLATE result in two new algorithms, GSCA and GSMA.

• GSCA/GSMA can construct a better conlitron/multiconlitron.

摘要

•Multiconlitron is a general piecewise linear classifier as a union of conlitrons.•A new technique is proposed for improving the performance of a conlitron/multiconlitron.•This technique consists of two basic operations, SQUEEZE and INFLATE.•SQUEEZE and INFLATE result in two new algorithms, GSCA and GSMA.•GSCA/GSMA can construct a better conlitron/multiconlitron.

论文关键词：Conlitron,Multiconlitron,Piecewise linear classifiers,Piecewise linear learning,Support vector machines

论文评审过程：Received 9 September 2013, Revised 20 February 2014, Accepted 31 March 2014, Available online 13 April 2014.

论文官网地址：https://doi.org/10.1016/j.knosys.2014.03.024