A hybrid model to infer US-Japan trade relations

摘要

In this paper, we propose a new neural model in which the information maximization and error minimization components are combined. Since information is maximized, information is compressed into networks in explicit ways, which enables us to discover the salient features in input patterns. We applied this method to a problem of US-Japan trade relations. Experimental results confirmed that, due to the maximized information in competitive units, easily interpretable internal representations can be obtained.