Multimodal Brain-Computer Interface Grand Challenge: EEG-fNIRS based Handwriting-Trajectory Classification

Badong Chen, Ziyu Jia, Roger Mark

Keywords: Brain-Computer Interface, EEG, fNIRS, Multimodal Learning, Imagined Handwriting, Classification

Abstract: Brain-computer interface (BCI) enables direct human--computer interaction by decoding neural activity into machine commands, with broad potential in communication and rehabilitation. In particular, non-invasive BCIs are attractive due to their low risk and cost-effectiveness, and motor imagery (MI) has become a widely studied paradigm because MI-based training can support stroke. Beyond classical limb MI aims to infer imagined writing movements and thus offers a natural route towards brain-to-text communication for people with severe paralysis. We propose an ACM Multimedia Grand Challenge on multimodal Brain-computer interfaces (BCIs) that targets a single, well-defined task: four-class classification of imagined logographic handwriting using synchronized electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS). The challenge is built on a multi-session EEG--fNIRS benchmark with aligned trial annotations and standardized preprocessing, enabling reproducible comparison of temporal modeling and multimodal fusion methods under realistic session variability. We define one track with a unified model-submission format, and all evaluation is performed automatically on a hidden test set by the organizers. The leaderboard is ranked solely by overall classification accuracy (ACC). We commit to maintaining the public challenge website, starter kit, and evaluation infrastructure for at least three years.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 26