Classification model training method and device and classification method and device

Abstract

The invention provides a classification model training method and device and a classification method and device. The classification model training method comprises steps that the preset number of training image frames in training videos with labels are acquired; the target neural network is utilized to carry out characteristic learning of the training image frames, and characteristic vectors are extracted for the training image frames; the training videos represented by the extracted characteristic vectors are classified based on memory characteristic vectors corresponding to different classifications stored in the memory network to acquire the classification result of the training videos; according to the classification result of the training videos and the comparison result between the labels of the training videos, the target neural network is trained, the memory network has strong memory ability, a problem of poor memory ability of the neural network can be made up, regardless of the input sequence of any length, characteristics can be memorized by the memory network well, therefore, regardless of the input sequence, identification accuracy of a classification model is not affected by the poor memory capacity of the neural network.

Description

technical field [0001] The present application relates to the technical field of machine learning, in particular, to a classification model training method and device, and a classification method and device. Background technique [0002] Multi-layer feedback (Recurrent neural Network, RNN) neural network, also known as recurrent neural network, is an artificial neural network in which nodes are directional connected into a ring, which can use internal memory to process input sequences of any time sequence, and in video action prediction been widely applied. [0003] However, the memory ability of the RNN network is poor. When the input sequence is long, it cannot accurately classify the input sequence and cannot meet the use requirements for accuracy. Contents of the invention [0004] In view of this, the purpose of the embodiments of the present application is to provide a classification model training method and device, and a classification method and device, which can...

AI Technical Summary

Problems solved by technology

[0003] However, the memory ability of the RNN network is poor. When the input sequence is long, it cannot accurately classify the input sequence and cannot meet the use requirements for accuracy.

Images