A kernel based learning method for non-stationary two-player repeated games

作者：

Highlights：

• String kernel density estimation, capable of predicting the opponent’s actions.

• Kernel functions with positive-definite property.

• Long-term memory can be stored in a R-way Trie data structure.

• Linear computational complexity for calculating the kernel for memory sequences.

摘要

•String kernel density estimation, capable of predicting the opponent’s actions.•Kernel functions with positive-definite property.•Long-term memory can be stored in a R-way Trie data structure.•Linear computational complexity for calculating the kernel for memory sequences.

论文关键词：Game theory,Repeated games,Sequence prediction,String kernel

论文评审过程：Received 7 February 2020, Revised 19 March 2020, Accepted 24 March 2020, Available online 1 April 2020, Version of Record 16 April 2020.

论文官网地址：https://doi.org/10.1016/j.knosys.2020.105820