Self-supervised pretraining improves brain signal decoding accuracy quickly
Pretraining for Sample-Efficient Neural Interfaces
Machine Learning
Summary
Decoding brain signals to help people regain lost abilities usually needs lots of labeled data collected from each person. The authors show that training a neural network on unlabeled brain signal data from many people first can create useful representations that work across different individuals. Their method uses spatial information about brain locations to better understand varied brain signals. This reduces the labeled data needed for accurate decoding by more than 20 times when applied to new people.
What this means in practice
- •For medical device developers: Design neural decoders that require far fewer labeled brain signal trials when fitting devices for new patients.
- •For ai engineers in neurotechnology: Build cross-patient neural representations from large unlabeled datasets to improve brain signal decoding generalization.
Authors
Ben Tang, Zachary Spalding, Gregory B. Cogan
Abstract
Brain-computer interfaces (BCIs) decode neural activity to restore lost function. Typically, training a high-performance neural decoder requires a large labeled dataset to be collected from every new subject. One way to reduce the labeled data cost is self-supervised pretraining, which learns general neural representations from unlabeled recordings that accumulate across subjects. However, for intracranial electroencephalography (iEEG) recordings, self-supervised learning has been challenging due to differences in contact placement and neuroanatomy between subjects. We propose MAPA, an otherwise vanilla masked autoencoder with two spatial encodings, an anatomical region embedding and a relative positional encoding, which together enable it to learn neural representations that transfer to unseen subjects and across various tasks. MAPA sets a new state of the art across all three regimes of the Neuroprobe benchmark without fine-tuning: within-session, cross-session, and cross-subject. In the cross-subject regime, a linear probe on MAPA's features needs only ${\sim}164$ labeled trials to reach the accuracy that takes 3,500 without pretraining. Our results show that self-supervised pretraining can scale across heterogeneous iEEG recordings and reduce the labeled data needed for accurate decoding in new subjects.