Portada

FEATURE OPTIMIZATION OF MOTOR IMAGERY EEG CLASSIFICATION USI IBD

LAP LAMBERT ACADEMIC PUBLISHIN
12 / 2024
9786205492451
Inglés

Sinopsis

Brain-computer interfaces (BCIs) hold great promise in biomedical engineering, particularly for diagnosing critical diseases. Motor imagery (MI) EEG classification, a key BCI process, faces challenges due to the complexity and non-stationary nature of EEG signals. These signals, recorded via electrodes, are digitized and analyzed using feature extraction techniques like FFT, STFT, CSP, and wavelet transforms, with wavelet transform being the most effective.This study proposes a deep neural network-based classification algorithm with teacher-learning-based optimization for feature refinement. Tested on a standard BCI dataset in MATLAB, the algorithm surpasses Bayesian and ensemble machine learning classifiers, enhancing classification accuracy and BCI system performance.

PVP
50,92