Reproduce “EEG2MCU: Sensing-to-Execution Co-Design for Wearable EEG Inference” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Shoma Yukawa, Meghna Roychowdhury, Arnab Sanyal, Chengwei Zhou, Shreyas Sen, Gourav Datta

EEG2MCU reduces wearable inference costs by selecting task-relevant electrodes and streaming only the current first-layer input tile, avoiding a fully materialized high-channel representation. On three-class SEED, its saliency-guided 19-electrode montage reached 73.4% accuracy versus 61.9% for the standard 19-electrode montage. The design reduced peak SRAM 11.38× versus an unfused 62-electrode baseline; results are limited to evaluated models and conditions.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.