Cross Task Neural Architecture Search for EEG Signal Classifications

TL;DR

This paper introduces a novel neural architecture search framework for EEG signal classification that automatically designs task-specific network structures, leading to improved accuracy across multiple EEG recognition tasks.

Contribution

It proposes the first cross-task neural architecture search framework for EEG, enabling automatic structure design and analysis of cross-task and cross-subject variations.

Findings

Achieved state-of-the-art performance on Motor Imagery and Emotion recognition tasks.

Demonstrated the effectiveness of automated architecture search over manual design.

Provided insights into structure differences across EEG tasks and subjects.

Abstract

Electroencephalograms (EEGs) are brain dynamics measured outside the brain, which have been widely utilized in non-invasive brain-computer interface applications. Recently, various neural network approaches have been proposed to improve the accuracy of EEG signal recognition. However, these approaches severely rely on manually designed network structures for different tasks which generally are not sharing the same empirical design cross-task-wise. In this paper, we propose a cross-task neural architecture search (CTNAS-EEG) framework for EEG signal recognition, which can automatically design the network structure across tasks and improve the recognition accuracy of EEG signals. Specifically, a compatible search space for cross-task searching and an efficient constrained searching method is proposed to overcome challenges brought by EEG signals. By unifying structure search on different EEG tasks, this work is the first to explore and analyze the searched structure difference cross-task-wise. Moreover, by introducing architecture search, this work is the first to analyze model performance by customizing model structure for each human subject. Detailed experimental results suggest that the proposed CTNAS-EEG could reach state-of-the-art performance on different EEG tasks, such as Motor Imagery (MI) and Emotion recognition. Extensive experiments and detailed analysis are provided as a good reference for follow-up researchers.