一种高速神经网络HS-K-WTA的算法研究

A High-Speed Neural Network HS-K-WTA Algorithm

该文提出一种新的K -Winners-Take -All神经网络 :High -Speed -K -Winners-Take-All(HS -K -WTA)。HS -K -WTA是以竞争学习算法为基础。HS -K -WTA能够从一个模式集中 ,识别出K个较大的元素 ,或较小的元素。该文给出HS -K -WTA算法及算法复杂性的分析。分析结果显示HS -K-WTA要比Winstrons更快。

A new K Winners Take All neural network, High Speed Winners Take All (HS K WTA) is presented.Based on the competitive study algorithm,HS K WTA can identify the K larger elements (or K smaller ones) in a pattern set. This paper analyzes the HS K WTA algorithm and its complexity. The results show that HS K WTA operates more quickly than Winstrons.