PC-SSL: A Predictive Coding-Based Self-Supervised Learning Framework for EEG Emotion Recognition

Tue, 28 Apr 2026 00:00:00 +0000

RML at ICASSP 2026: PC-SSL for EEG Emotion Recognition

Tue, 28 Apr 2026 00:00:00 +0000

Niki Sheibani and Dr. Naimul Khan will present their paper “PC-SSL: A Predictive Coding-Based Self-Supervised Learning Framework for EEG Emotion Recognition” at ICASSP 2026.

PC-SSL addresses the challenge of limited labeled EEG data by leveraging unlabeled recordings through a predictive coding framework. The model uses a convolutional encoder with band-wise and channel-wise attention to extract spatio-spectral features from EEG signals, then fine-tunes on annotated data. On the SEED-IV and SEED-V benchmark datasets, PC-SSL achieves 84.48% and 92.39% accuracy — improving over previous state-of-the-art by ~18% and ~16% respectively.

Code is available at github.com/Niki-sh/PC-SSL. Read the paper on IEEE Xplore.

Self-Supervised Learning | TMU Multimedia Research Laboratory

PC-SSL: A Predictive Coding-Based Self-Supervised Learning Framework for EEG Emotion Recognition

RML at ICASSP 2026: PC-SSL for EEG Emotion Recognition