Search papers, labs, and topics across Lattice.
We introduce MEG-Mamba, a generative foundation model for magnetoencephalography (MEG) data utilizing the Mamba architecture, trained to predict future brain activity from parcellated source-reconstructed MEG signals. This approach achieves superior generative fidelity and significantly faster pre-training with longer context modeling compared to a transformer-based MEG-GPT. Furthermore, lightweight stimulus conditioning enables the generation of task-evoked responses, showcasing Mamba's potential for scalable neural foundation models.
Mamba-based foundation models for MEG outperform transformers in generative fidelity and training efficiency, unlocking longer context modeling of brain activity.
Magnetoencephalography (MEG) is an imaging technique that offers a non-invasive, millisecond-resolution view of human brain activity. The increasing availability of MEG data presents an opportunity to take advantage of a recent advance in artificial intelligence, namely self-supervised foundation models. Existing foundation models for MEG (and electroencephalography) have been built on a transformer architecture. Here, we introduce MEG-Mamba: a generative foundation model for neural activity (source reconstructed, parcellated MEG) built on a Mamba architecture. MEG-Mamba is trained to predict the next token of a discretised MEG signal, conditioned on a brain region and recording session. MEG-Mamba surpasses the generative fidelity of a transformer-based alternative (MEG-GPT) while pre-training in less time (22 vs 400 GPU-hours) and modelling a longer context (4 vs 0.32 s). We evaluate MEG-Mamba by examining the spatio-spectral characteristics of the data it generates and by interpreting its learned embeddings. Furthermore, we demonstrate that lightweight stimulus conditioning (LoRA) can be used to generate realistic task-evoked responses that are not included in pre-training. Our results suggest that Mamba is a promising backbone for scaling neural foundation models.
