EEG Channels Selection Based on BiLSTM and NSGAII

Emotion recognition work for EEG signals has become one of the most important measures for researchers to explore human-computer interaction work. However, the traditional emotion recognition approach utilizes all EEG channels which may lead to increased computational degree as well as un-wanted interfering information affecting the accuracy. And it is not suitable for all emotion recognition work. In this paper, we propose an EEG selection framework based on bidirectional long short-term memory network (BiLSTM) and non-dominated sorting genetic algorithm-II (NSGAII) to select the optimal set of EEG channels for emotion recognition. The EEG data is first identified using BiLSTM, followed by optimization of the results using NSGAII, and continuous iteration to arrive at the optimal channel set. The experiments were conducted using the publicly available dataset DEAP, and the experimental results show that the method reduces the number of channels and maintains a high emotion recognition accuracy.