Scaling EEG models improves brain signal analysis accuracy
LaBraM at Scale: Scaling Data and Models for EEG Pretraining
Machine LearningArtificial Intelligence
Summary
Analyzing brain signals recorded by EEG devices is tricky because the signals vary a lot. The researchers extended an existing method called LaBraM by using much more data and bigger models, making their system better at understanding EEG signals for many tasks. They found that bigger models trained with more data often perform better, but bigger isn’t always better for every kind of adaptation. Their work shows how making models and data larger can improve brain signal analysis, but the match between training and later use is important.
What this means in practice
- •For medical device developers: Improve brain signal interpretation across diverse clinical EEG tasks using larger pretrained EEG models built and evaluated here.$Commercial implications: Enables advanced EEG-based diagnostic tools with higher accuracy applicable across multiple brain monitoring uses.
- •For machine learning engineers: Employ scaled-up pretrained EEG encoders and training methods to enhance EEG-based prediction models in various applications.
Authors
Jun-Yu Pan, Ruicheng Yin, Bao-Liang Lu, Dongsheng Li, Wei-Long Zheng, Yansen Wang
Abstract
LaBraM, an early EEG foundation model, demonstrated that large-scale EEG pretraining can yield strong representations for diverse downstream tasks. Building on this foundation, we expand pretraining data and model capacity while retaining its core encoder and objective. We train ten encoders with 5.8 million to 1.97 billion parameters on three EEG collections spanning 25,067 to 79,910 hours, adapting signal preparation and training to heterogeneous recordings. Across 23 tasks, Atom-L* reaches 59.05% mean balanced accuracy at its best evaluated full fine-tuning (Full-FT) checkpoint, 3.61 percentage points above the strongest baseline EEG foundation model. Quark-G leads linear probing (LP) at 54.33%, a 1.24-point margin. Both also lead average task rank. At the last evaluated checkpoints, larger encoders improve Full-FT more consistently than LP, and stronger LP does not guarantee stronger Full-FT. Preserving patch structure improves LP over average pooling. Data-fraction experiments suggest capacity-dependent saturation, while checkpoint trajectories show that downstream performance can decline as pretraining loss falls. Joint classification with a shared Qwen3 readout reaches 50.50% with Molecule-H and Qwen3-8B; increasing language-model size provides no consistent gain. These results demonstrate the value of scaling LaBraM and the importance of matching pretraining scale to downstream adaptation.