Motor Imagery Decoding Using Ensemble Curriculum Learning and Collaborative Training

TL;DR

This paper tackles cross-subject motor imagery decoding from EEG by introducing a two-stage ensemble that uses multiple feature extractors and a shared classifier. It combines a curriculum learning strategy that forces each extractor to specialize to a subset of subjects with an intra-ensemble distillation loss that promotes collaboration among models, achieving strong cross-subject generalization with relatively few parameters. Evaluations on PhysioNet and OpenBMI show state-of-the-art 5-fold CV and LOSO performance, notably 86.36% on PhysioNet (K=7) and 79.73% on OpenBMI (K=3) without test-data adaptation. The approach advances calibration-free BCI by leveraging diverse representations and cross-model knowledge exchange, and the authors provide public code for reproducibility.

Abstract

In this work, we study the problem of cross-subject motor imagery (MI) decoding from electroencephalography (EEG) data. Multi-subject EEG datasets present several kinds of domain shifts due to various inter-individual differences (e.g. brain anatomy, personality and cognitive profile). These domain shifts render multi-subject training a challenging task and also impede robust cross-subject generalization. Inspired by the importance of domain generalization techniques for tackling such issues, we propose a two-stage model ensemble architecture built with multiple feature extractors (first stage) and a shared classifier (second stage), which we train end-to-end with two novel loss terms. The first loss applies curriculum learning, forcing each feature extractor to specialize to a subset of the training subjects and promoting feature diversity. The second loss is an intra-ensemble distillation objective that allows collaborative exchange of knowledge between the models of the ensemble. We compare our method against several state-of-the-art techniques, conducting subject-independent experiments on two large MI datasets, namely PhysioNet and OpenBMI. Our algorithm outperforms all of the methods in both 5-fold cross-validation and leave-one-subject-out evaluation settings, using a substantially lower number of trainable parameters. We demonstrate that our model ensembling approach combining the powers of curriculum learning and collaborative training, leads to high learning capacity and robust performance. Our work addresses the issue of domain shifts in multi-subject EEG datasets, paving the way for calibration-free brain-computer interfaces. We make our code publicly available at: https://github.com/gzoumpourlis/Ensemble-MI

Figures (1)

• Figure 1: Our proposed architecture has $\mathrm{K}$ first stage feature extractors and a shared classifier in the second stage. Each feature extractor is trained: i) on an ensemble curriculum learning objective ($\mathcal{L}_{\mathrm{subj}}$) and ii) on a knowledge distillation objective ($\mathcal{L}_{\mathrm{distill}}$).