A Comparative Study of Subject-Dependent and Subject-Independent Strategies for EEG-Based Emotion Recognition using LSTM Network

Abstract

This paper addresses the problem of EEG-based emotion recognition and classification and investigates the performance of classifiers for subject-independent and subject-dependent models separately. The results are compared with other classifiers and also with existing work in the concerned domain as well. We perform the experiments on the publicly available DEAP dataset with band power as the feature and classification accuracies are found pertaining to the widely accepted Valence-Arousal Model. The best results were reported by the LSTM model in case of the subject-dependent model with accuracies of 94.69% and 93.13% on valence and arousal scales respectively. SVM performed the best for the subject-independent model with accuracies of 72.19% on valence scale and 71.25% on arousal scale.