Deep neural network learning method and apparatus, and category-independent sub-network learning apparatus

Abstract

Provided is a DNN learning method that can reduce DNN learning time using data belonging to a plurality of categories. The method includes the steps of training a language-independent sub-networkand language-dependent sub-networksandwith training data of Japanese and English. This step includes: a first step of training a DNN obtained by connecting neurons in an output layer of the sub-networkwith neurons in an input layer of sub-networkwith Japanese training data; a step of forming a DNN by connecting the sub-networkin place of the sub-networkto the sub-networkand training it with English data; repeating these steps alternately until all training data ends; and after completion, separating the first sub-networkfrom other sub-networks and storing it as a category-independent sub-network in a storage medium.