摘要 We extend Linkser's Infomax principle for feedfor

本頁只刊出中文翻譯與中文說明；英文原文請見下方原文連結。

原文連結

• SFI 頁面

論文資訊

• 類型：已發表論文
• 日期：2005

摘要

We extend Linkser's Infomax principle for feedforward 神經 網絡s to a measure for 隨機 interdependence that captures spatial and temporal signal properties in recurrent systems. This measure, 隨機 interaction, quantifies the Kullback-Leibler divergence of a 馬可夫 chain from a product of split chains for the single unit processes. For unconstrained 馬可夫 chains, the maximization of 隨機 interaction, also called Temporal Infomax, has been previously shown to result in almost deterministic dynamics. This letter considers Temporal Infomax on constrained 馬可夫 chains, where some of the units are clamped to prescribed 隨機 processes providing input to the system. Temporal Infomax in that case leads to finite state automata, either completely deterministic or weakly nondeterministic. Transitions between internal s

※ 此為已發表論文，全文需透過期刊付費取得