DWT- and Parametric t-SNE-based characteristic extracting method of motor imagery EEG(Electroencephalogram) signals

Abstract

The invention provides a DWT- and Parametric t-SNE-based characteristic extracting method of motor imagery EEG(Electroencephalogram) signals. First, effective time and frequency ranges of EEG characteristics are determined by using a Wigner-Ville distribution and power spectrum; the EEG signals in a specific time and frequency segment is subjected to three-layer discrete wavelet decomposition and statistical characteristic quantity including the average value, the energy average value, the mean square error and the like are calculated and are taken as the time frequency characteristic of the EEG signals; at the same time, a parameterization t-SNE algorithm is utilized for performing non-linear characteristic mapping on said wavelet coefficients and embedded coordinates corresponding to a low-dimensional space are taken as the non-linear characteristic; the two characteristics are standardized and a characteristic vector including both the time frequency information and the non-linear information of the EEG signals in the specific time frequency segment is obtained. According to the invention, EEG characteristics of compactness and completeness are obtained and a method for solving a problem of poor generalization performance of a traditional manifold learning algorithm in pattern classification application through fitting a multilayer forward propagation neural network to nonlinear mapping is proposed, so that accuracy of pattern classification of MI-EEG signals is improved further.

Description

technical field [0001] The invention is an EEG signal processing technology, which is specifically applied to the extraction of motor imagery EEG signal features in a Brain-Computer Interface (BCI) system, using discrete wavelet transform (DiscreteWaveletTransform, DWT) and parameterized t The method of combining Parametric-Distributed Stochastic NeighborEmbedding (Parametric-SNE) is used for feature extraction and fusion of motor imagery EEG signals. Background technique [0002] Motor Imagery Electroencephalography (MI-EEG) contains the imaginer's willingness to move and rich neurophysiological information. It has attracted much attention in the research fields of brain cognition and brain application. The correct interpretation and accurate extraction of MI-EEG Feature information is the key to its successful application. [0003] In view of the characteristics of individual differences, nonlinearity, non-stationary and time-sensitive changes in MI-EEG signals, the wavel...

AI Technical Summary

Problems solved by technology

[0005] However, in practical applications, we found that the existing popular learning methods have the following shortcomings: (1) Sensitive to data noise, for EEG signals with significant non-stationarity, randomness and noise, use manifold learning methods for feature extraction or When feature dimensionality is reduced, it is easy to destroy the low-dimensional embedding structure, thereby affecting the quality of features; (2) EEG signals have significant time-frequency distribution characteristics and nonlinear characteristics, making it difficult to fully obtain their essential features using only ML algorithms, and the feature vector cannot be guaranteed compactness and completeness, and even cause feature information redundancy and feature mismatch problems; (3) traditional manifold learning methods can only perform data dimensionality reduction on a given data set, and cannot produce an explicit The mapping relationship from high-dimensional observation space to low-dimensional embedding space is not conducive to the generalization ability of manifold learning for out-of-sample data

Therefore, the current data dimensionality reduction methods represented by manifold learning have low pattern classification accuracy in MI-EEG signals.

Images