An identification method and system based on deep learning neural network

Abstract

The invention discloses an identity recognition method and system based on a deep learning neural network. The method is as follows: first obtain N types of input data; then initialize corresponding N types of deep learning neural networks; then train N types of deep learning neural networks; Then sort the categories according to the collection cost from low to high; then initialize i, the optimal output label L, and the maximum similarity relative ratio U; then calculate the Ti class test output label Li, and judge whether L is the same as Li; then obtain Ti Class similarity relative ratio Ui, and according to the result of the previous step, it is judged whether the identification fails or the next step is performed; then L and U are updated; finally, according to the relationship between U and the preset maximum similarity relative ratio c, and the relationship between i and N, It is judged whether the identification is successful, and whether it is necessary to add 1 to i to continue the loop execution. In the present invention, low-cost test data is selected preferentially, and test input types are gradually increased, so as to obtain the highest discrimination accuracy rate under the condition of the lowest cost.

Description

technical field [0001] The present invention relates to an identification method and system, in particular to an identification method and system based on a deep learning neural network. Background technique [0002] In the prior art, the deep learning technology can obtain the output label through the input data (such as obtaining the ID card number of the person through the avatar, or obtaining the ID card number of the person through the voice). The labeled data (for example, the avatar with the ID number, and the voice with the ID number) is supervised training (the data sample is used as the input of the deep learning neural network, and the label is used as the output of the deep learning neural network). [0003] However, since the accuracy rate of the deep learning neural network cannot reach 100%, the ID number output after inputting a profile picture into the deep learning neural network may be the ID number of another person with a similar profile picture. The ou...

AI Technical Summary

Problems solved by technology

[0003] However, since the accuracy rate of the deep learning neural network cannot reach 100%, the ID number output after inputting a profile picture into the deep learning neural network may be the ID number of another person with a similar profile picture. The output ID number may be the ID number of another person with a similar voice

When performing user identification, in order to improve the accuracy of identification, it is generally necessary to collect multiple types of user data for comprehensive judgment, but this will increase the cost of collection, and the more types of data collected, the more inconvenience it will cause to users. Convenience, and greatly save the user's time

Therefore, it is necessary to obtain accurate judgments through as few test categories as possible, but the existing deep learning neural network technology cannot realize comprehensive discrimination through multi-type deep learning neural networks, and it is also impossible to select categories

Images