Calibration free meta learning based approach for subject independent EEG emotion recognition

• Published in: Biomedical signal processing and control Vol. 72; p. 103289
• Main Authors: Bhosale, Swapnil, Chakraborty, Rupayan, Kopparapu, Sunil Kumar
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.02.2022
• Subjects: EEG emotion detection; Episodic training; Few-shot learning; Support sampling; Zero calibration setup; EEG emotion detection; Support sampling; Zero calibration setup; Few-shot learning; Episodic training
• ISSN: 1746-8094; 1746-8108
• DOI: 10.1016/j.bspc.2021.103289

•Proposed a few-shot adaptation for EEG based Emotion Recognition which limits the number of samples required for calibration samples to a minimum.•Devised and demonstrated the efficacy of three sampling strategies for support sets during our training for scenarios when reference samples from the subject under consideration may or may not be present, which is novel.•Employed a 1-D EEG signal transformation into a 3-D matrix representation such that the spatial correlation among adjacent EEG electrodes is also learnt in addition to high-level features of individual electrodes.•Proposed a 3-D convolutional recurrent embedding architecture to extract the temporal relations from the spatially convolved features of individual electrodes, which is also novel.•The combination of our best sampling strategies in subject dependent and subject independent setups outperforms existing approaches using a cross-subject evaluation setup. Brain Computer Interfaces (BCI) detect changes in the electrical activity of brain which could be applied in use-cases like environmental control, neuro-rehabilitation etc. Prior to the actual usage, the subject has to undergo a lengthy calibration phase and hence prohibits an optimal plug-and-play experience. To quantify the minimum number of samples required for calibration, we propose a few-shot adaptation to the task of recognizing emotion from Electroencephalography (EEG) signals, without requiring any fine-tuning of the pre-trained classification models for every user. Our experiments illustrate the usefulness of various sampling strategies based on the presence or absence of subject dependent and subject independent reference samples during training. In comparison with the existing state-of-the-art model, which is trained in a supervised manner, our approach with only 20 reference samples from subjects under consideration (unseen during training) shows an absolute improvement of 8.56% and 7.53% in accuracy on emotion classification in valence and arousal space, respectively, without any re-training using the samples from the unseen subjects. Moreover, when tested in a zero calibration setup (when reference samples are taken from subjects other than the subject under consideration), our system improves the accuracy over the supervised model by 2.02% and 0.61% for emotion classification in valence and arousal space, respectively.