Identity authentication method and authentication system based on EEG features

Abstract

The invention discloses an identity authentication method and an authentication system based on electroencephalogram features, wherein the method includes five steps of model training, password setting, alpha wave testing, electroencephalogram feature extraction and classification, and model updating. Among them, the brain wave feature extraction and classification method is composed of three parts: layer-by-layer pre-training, network fine-tuning, and classification. The present invention first utilizes the characteristic that α waves fluctuate greatly when a person tells a lie to carry out the first re-authentication on the tester, improves the existing method at the same time, establishes a feature extraction method based on a deep belief network, and uses the immune leapfrog algorithm to It is optimized to perform feature extraction and classification on the tester's brain waves when observing the password pattern, so as to perform the second verification. The two-layer verification can not only ensure the safety of users, but the improved deep belief network can also improve the existing recognition accuracy, reduce the error rate, avoid a large number of false alarms, and bring a good experience to users.

Description

technical field [0001] The present invention relates to the field of Internet of Things technology and information security technology, in particular to an identity authentication method and authentication system based on EEG features. Background technique [0002] Traditional identity authentication methods include voice authentication, face recognition, password authentication, fingerprint authentication, and iris authentication. These traditional identity authentication methods have many disadvantages: some of them are easy to be breached, such as password leaks, fingerprints stolen, etc.; Therefore, an identity authentication system that is difficult to break, difficult to leak, and has high recognition accuracy came into being-the brain wave identity authentication system. Experimental studies have shown that EEG characteristics are unique to each person. Even when the same person is facing different events or in different mental states, EEG characteristics will be ver...

AI Technical Summary

Problems solved by technology

There are many shortcomings in these traditional identity authentication methods: some of them are easy to be broken, such as password leakage, fingerprints stolen, etc.;

[0003] Most of the existing brainwave identity authentication systems use convolutional neural networks or deep belief networks as feature extraction and classification methods. However, the implementation of convolutional neural networks is relatively complicated, and the cost of specific instantiation is high, so it is not suitable for promotion. , and due to the separate training of different tasks, the training time is relatively long, so users may not be able to use it in time; while the traditional deep belief network sometimes falls into local optimum due to the random assignment of weights between layers and produces "premature " phenomenon, so the present invention proposes to use the deep belief network optimized by the immune leapfrog algorithm to extract and classify brain wave features, so that not only can jump out of the local optimal solution, but also speed up the calculation

Images