Skip to content

Enhancing EEG-Based Emotion Recognition Using Sparse Dynamic Graph CNN with ℓ2,1-Norm

Published

Venue: IEEE Sensors Journal
Author role: Second author
Year: 2025

Research approach

Applied row-wise ℓ₂,₁ sparsity and proximal updates to a learned EEG electrode graph.

Sparse DGCNN architecture and sparsity regularization — Sparse DGCNN architecture and sparsity regularization
View full-size figure

Graph edge-density analysis for Sparse DGCNN — Graph edge-density analysis for Sparse DGCNN
View full-size figure

Results

Subject-independent accuracy with DE features: 78.36 ± 9.53% on SEED and 47.67 ± 6.67% on SEED-IV.

All publications