EEG-X: Toward Device-Agnostic and Noise-Robust Foundation Model for EEG
Organizations: Emotiv Research, Melbourne, Australia · Emotiv Research, Sydney, Australia · Monash University, Melbourne, Australia
Abstract
Foundation models for EEG analysis are still in their infancy, limited by two key challenges: (1) variability across datasets caused by differences in recording devices and electrode configurations, and (2) the low signal-to-noise ratio (SNR) of EEG, where neural activity is often buried under artifacts and non-brain sources. To address these challenges, we present EEG-X, a device-agnostic and noise-robust foundation model for EEG representation learning. EEG-X introduces a location-based channel embedding that enables robust transfer across heterogeneous EEG devices and electrode layouts. To improve robustness against noise, EEG-X employs a novel noise-aware masking-reconstruction strategy that reconstructs artifact-cleaned signals rather than raw noisy EEG and introduces a novel Dictionary-based Convolutional Transformation (DiCT) that improves reconstruction-based pretraining by comparing signals in a structured feature space instead of directly in the raw signal space. Experiments across datasets collected from diverse EEG devices show that EEG-X consistently outperforms state-of-the-art methods across multiple downstream tasks and demonstrates strong cross-domain generalization when pretraining and downstream datasets differ in electrode layouts and recording configurations paving the way towards foundation models. The models and code are available at https://github.com/Emotiv/EEG-X.
Figures & tables
| Category | Datasets | Channel | Length | #Samples | Task |
|---|---|---|---|---|---|
| In-Domain (Pre-training) | DREAMER | 14 | 2s | 170,246 | Emotion Detection |
| STEW | 14 | 2s | 56,928 | Mental Workload Classification | |
| TUAB | 19 | 10s | 409,455 | Abnormal EEG Classification | |
| TUEV_binary | 19 | 2s | 98,684 | Epileptiform Classification | |
| Cross-Domain (Downstream) | Crowdsourced | 14 | 2s | 12,296 | Eyes Open/Close Detection |
| BCICIV_2A | 22 | 4s | 5,184 | Motor Imagery |
| Model | STEW | DREAMER | TUAB | TUEV_binary | ||||
|---|---|---|---|---|---|---|---|---|
| Acc | AUROC | Acc | AUROC | Acc | AUROC | Acc | AUROC | |
| MAEEG | 75.98 2.87 | 81.73 2.95 | 52.36 3.12 | 54.45 2.98 | 77.56 1.45 | 86.56 1.67 | 85.75 2.03 | 91.36 1.89 |
| BioT | 75.79 1.12 | 86.48 1.45 | 52.85 1.34 | 54.19 1.23 | 79.21 1.56 | 87.42 1.78 | 86.28 1.43 | 93.06 1.67 |
| LaBraM | 75.26 1.35 | 84.37 1.67 | 55.85 1.45 | 55.34 1.23 | 80.61 1.54 | 88.77 1.76 | 85.43 1.48 | 90.10 1.59 |
| EEGPT | 77.65 2.95 | 85.98 3.01 | 54.11 3.12 | 54.91 2.87 | 78.91 1.49 | 87.16 1.85 | 84.05 2.03 | 90.26 1.78 |
| EEG2Rep | 76.71 1.23 | 85.89 1.45 | 57.09 1.32 | 56.35 1.21 | 80.21 1.54 | 88.43 1.67 | 89.83 1.38 | 95.57 1.47 |
| Model | Crowdsourced | BCICIV_2A | SSVEP | PhysioNetMI | ||||
|---|---|---|---|---|---|---|---|---|
| Acc | AUROC | B-Acc | Weighted F1 | B-Acc | Weighted F1 | B-Acc | Weighted F1 | |
| BioT | 77.76 1.12 | 83.62 1.34 | 27.26 1.05 | 25.26 1.23 | 29.92 1.41 | 27.27 1.36 | 48.51 2.70 | 59.03 2.44 |
| LaBraM | 77.28 1.25 | 89.83 1.47 | 32.44 1.13 | 27.37 1.21 | 29.69 1.28 | 27.41 1.19 | 47.41 1.60 | 57.92 1.48 |
| EEGPT | 79.55 1.36 | 87.79 1.52 | 30.07 1.17 | 24.68 1.08 | 26.06 1.11 | 27.40 1.07 | 46.96 2.10 | 57.37 1.95 |
| CBraMod | 78.41 1.31 | 88.73 1.45 | 30.07 1.12 | 24.68 1.09 | 28.24 1.34 | 27.79 1.25 | 47.88 3.10 | 58.26 2.87 |
| REVE | 83.92 1.29 | 90.72 1.38 | 35.91 1.11 | 33.88 1.16 | 31.18 1.35 | 27.94 1.27 | 50.33 1.43 | 60.88 1.31 |
| In-Domain | |||||
|---|---|---|---|---|---|
| Component | STEW | DREAMER | TUAB | TUEV_binary | Average |
| Without channel embedding | 79.44 | 56.17 | 79.94 | 90.46 | 76.50 ( 1.16) |
| Learnable channel embedding | 80.07 | 57.12 | 80.94 | 90.14 | 77.07 ( 0.59) |
| Without DiCT | 76.91 | 54.36 | 78.88 | 88.01 | 74.54 ( 3.12) |
| Raw reconstruction | 77.90 | 55.41 | 79.41 | 88.22 | 75.24 ( 2.42) |
| Full EEG-X | 80.32 | 58.14 | 81.02 | 91.15 | 77.66 |
| Setting | STEW | DREAMER | TUAB | TUEV_binary |
|---|---|---|---|---|
| Supervised (Raw input) | 75.12 2.64 | 55.69 1.34 | 78.12 0.98 | 88.24 2.09 |
| Supervised (rASR input) | 73.68 1.74 | 56.41 1.04 | 78.32 1.11 | 83.36 1.68 |
| Self-Supervised (Raw input with rASR target) | 80.32 1.21 | 58.14 1.35 | 81.02 1.48 | 91.15 1.41 |
Appendix figures & tables8 assets
Supplementary material from the paper’s appendix.
Appendix
| Target | MSE (Time) | MSE (DiCT) |
|---|---|---|
| Low-frequency dominant | 47.55 | 11.44 |
| Mid/High-frequency dominant | 95.75 | 4.26 |
| Phase-shifted signal | 126.79 | 0.02 |
| Groups ( ) | Kernels ( ) | STEW Acc (%) | DREAMER Acc (%) |
|---|---|---|---|
| 4 | 16 | 80.15 | 58.00 |
| 8 | 32 | 80.32 | 58.14 |
| 64 | 16 | 80.28 | 58.10 |
| 8 | 64 | 80.30 | 58.12 |
| Setting | REVE | ST-EEGFormer | EEG2Rep | CBraMod | LaBraM | EEGPT | BioT | MAEEG |
|---|---|---|---|---|---|---|---|---|
| In-domain | 0.5324 | 0.1700 | 0.0264 | 0.0056 | 0.0036 | 0.0036 | 0.0036 | 0.0036 |
| Emb. dim | Enc. params | Total params | Inference GFLOPs / window | Pretrain TFLOPs / epoch | STEW | DREAMER |
|---|---|---|---|---|---|---|
| 64 | 0.40M | 0.65M | 0.114 | 44.97 | 80.32 | 58.14 |
| 128 | 1.59M | 2.59M | 0.405 | 72.70 | 80.70 | 58.68 |
| 256 | 6.33M | 10.28M | 1.515 | 179.87 | 80.75 | 59.12 |
| 512 | 25.2M | 41.01M | 5.848 | 601.05 | 80.78 | 60.38 |
EEG Foundation Models
Electroencephalography (EEG) foundation models learn reusable representations of brain signals for transfer across people, tasks, and recording setups.
See all 120 papers