Deep self-coding neural network-based electroencephalogram EEG signal feature extraction method

Abstract

The invention discloses a deep self-coding neural network-based electroencephalogram EGG signal feature extraction method. The method comprises the following steps of: S1, designing an EGG data acquisition test scheme for color recognition: and S2, designing transition pictures which comprise three test pictures and three full-black pictures in each period, wherein the test picture consume a timet1, the transition pictures consume a time t2, the test pictures in each period has three colors such as red, green and blue, and the sequences of red, green and blue are random. According to the method, the self-coding neural network does not require whether signals have steady and random conditions or not, and through repeated iterative training, the self-coding neural network is capable of learning features in the signals and restoring EGG signals approximately consistent with the original signals after repeated compression and decompression, so that the problem of determining orders in EGGsignal features by using an AR model is avoided, and features extracted by using the deep self-coding neural network can obtain satisfied results in the aspect of color recognition.

Description

technical field [0001] The invention relates to the technical field of EEG feature extraction and EEG color recognition, in particular to a method for extracting EEG features of EEG signals based on a deep self-encoding neural network. Background technique [0002] Related research in the field of EEG can be traced back to the end of the 20th century. Poulos M (1999) used FFT to extract EEG signal features, and used LVQ neural network for identity recognition and classification; Poulos M (2002) used linear AR model to extract EEG signal features; Mohammadi G(2006) used the linear AR model to extract EEG signal features, and used the competitive neural network to classify; Palaniappan R(2007) used the power of EEG signals as features; HTouyama(2009) used PCA to reduce the dimensionality of EEG signals, Use dimensionality-reduced EEG data as features; Tangkraingkij P (2009) used ICA to extract EEG signal features; La Rocca D (2012) used AR model to extract EEG signal features;...

AI Technical Summary

Problems solved by technology

[0003] Usually collected EEG data is a multidimensional time series set, that is, a data set composed of time series on each scalp electrode, so EEG is a high-dimensional data set, and in the question of which scalp electrode data to study , in previous EEG papers, there are: (1) Treat each scalp electrode as independent, perform feature extraction on the data of each scalp electrode, and finally average the experimental results of each scalp electrode, but this This approach ignores the possible mutual influence between each scalp electrode; (2) choose to combine multiple scalp electrodes according to experience or exhaustive method, this approach makes up for the defects of method (1), but in practice In the application process, the time required for this method is much less than (1), and the electrode combination selected by experience lacks certain objectivity and scientificity

Images