RECOGNITION APPARATUS BASED ON DEEP NEURAL NETWORK TRAINING APPARATUS AND METHODS THEREOF

摘要

The present invention relates to a deep neural network based recognition device, a training device and methods thereof. A deep neural network is obtained by inputting and training a training sample including a positive sample and a negative sample to an input layer of the deep neural network. The recognition device comprises a determination unit composed to determine that a sample to be recognized is a suspected abnormal sample, when a degree of reliability of a positive sample class is smaller than all predetermined threshold values in a classification result output by an output layer of the deep neural network. Therefore, the reliability of the classification result output by the deep neural network can be efficiently improved.