MEG-XL: Data-Efficient Brain-to-Text via Long-Context Pre-Training

TL;DR

MEG-XL introduces a long-context pre-training approach for brain-to-text interfaces, significantly improving data efficiency and transferability in neural decoding tasks.

Contribution

The paper presents MEG-XL, a model pre-trained with 2.5 minutes of neural data per sample, enabling better generalization and performance with less training data.

Findings

MEG-XL matches supervised performance with only 1 hour of data compared to 50 hours.

Pre-training with longer neural context improves transfer to word decoding.

Long-context pre-training exploits extended neural information often discarded by other methods.

Abstract

Clinical brain-to-text interfaces are designed for paralysed patients who cannot provide extensive training recordings. Pre-training improves data-efficient generalisation by learning statistical priors across subjects, but these priors critically depend on context. While natural speech might unfold gradually over minutes, most methods pre-train with only a few seconds of context. Thus, we propose MEG-XL, a model pre-trained with 2.5 minutes of MEG context per sample, 5-300x longer than prior work, and equivalent to 191k tokens, capturing extended neural context. Fine-tuning on the task of word decoding from brain data, MEG-XL matches supervised performance with a fraction of the data (e.g. 1hr vs 50hrs) and outperforms brain foundation models. We find that models pre-trained with longer contexts learn representations that transfer better to word decoding. Our results indicate that long-context pre-training helps exploit extended neural context that other methods unnecessarily discard. Code, model weights, and instructions are available at https://github.com/neural-processing-lab/MEG-XL .