A P300 detection method based on CNN-LSTM network

Abstract

The invention discloses a P300 detection method based on a CNN-LSTM network which comprises the following steps: 1) designing a P300 character speller, collecting EEG data and determining a training set and a test set; 2) preprocess that training set and the t set; 3) Establish the algorithm model of combining CNN and LSTM; 4) train that model parameters by use a training set; 5) Testing model character recognition accuracy, P300 wave recognition accuracy, recall, accuracy, and F-measure. The invention combines CNN and LSTM neural networks to establish an algorithm model, which not only takesinto account the time and space characteristics, but also solves the problem of gradient disappearance or explosion in a simple RNN algorithm, and further improves the accuracy, and is a feasible method for detecting P300 signals.

Description

technical field [0001] The invention relates to the technical field of EEG signal detection, in particular to a P300 detection method based on a CNN-LSTM network. Background technique [0002] A brain-computer interface (BCI) is a direct connection pathway created between the human or animal brain and external devices. It can be used to analyze the hidden information in EEG signals. Research on brain-computer interfaces has a long history, dating back to the 1970s. Electroencephalography (electro-encephalography, EEG) is a non-invasive brain-computer interface method that collects brain bioelectricity directly from the scalp. It has high time accuracy and does not need to implant monitoring channels. It is the most commonly used method today. Brain-computer interface signals. Event-related potential (ERP) is a special brain evoked potential in EEG, which has a time-locked relationship with the occurrence of specific events. The P300 wave is one of the commonly used signa...

AI Technical Summary

Problems solved by technology

However, traditional machine learning algorithms need to manually acquire features, which take a lot of time and have poor generalization ability

Although the existing deep learning algorithms can use unsupervised or semi-supervised feature learning and hierarchical feature extraction to replace manual feature acquisition, the generalization ability and accuracy rate are slightly improved.

However, the accuracy of character recognition still needs to be further improved.

Images