Binary Classification of Normal vs Abnormal EEG Signals on TUAB
85.1Balanced AccuracyCLEF
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CLEFProtocol=Linear probing2026.05 | 85.1 | — | 93.9 | — | — | |
| KAST-BAR-LargeParameters=2.2B, Training Protocol=Multi-Task2026.05 | 83.2 | — | — | — | — | |
| KAST-BAR-LargeModel Parameter=2.2B, Task Protocol=Multi-Task2026.05 | 83.2 | 87.1 | 90.7 | — | — | |
| CodeBrain2025.06 | 82.94 | 91 | 90.3 | — | — | |
| CBraMod2025.06 | 82.89 | 92.58 | 92.27 | — | — | |
| CodeBrain (Excluding)variant=Excluding2025.06 | 82.88 | 90.61 | 90.12 | — | — | |
| Brain-OF Huge2026.02 | 82.87 | — | — | — | — | |
| LaBraM-HugeSize=369.8M2026.04 | 82.58 | 92.04 | 91.62 | — | — | |
| NeuroLMProtocol=Linear probing2026.05 | 82.5 | — | 90.8 | — | — | |
| CBraMod 100%Pretrain Dataset (Hours)=90002026.03 | 82.49 | — | — | 92.21 | 91.56 | |
| CBraModSize=69.3M2026.04 | 82.49 | 92.21 | 91.56 | — | — | |
| THD-BAR-HugeParameters=1.6B, Training Protocol=Multi-Task2026.05 | 82.2 | — | — | — | — | |
| THD-BAR-HugeModel Parameter=1.6B, Task Protocol=Multi-Task2026.05 | 82.2 | 84.8 | 88.6 | — | — | |
| Brain-OF Large2026.02 | 81.9 | — | — | — | — | |
| LaBraM-GBWAttBackbone foundation model=LaBraM, Attention mechanism variant=GBWAtt2026.06 | 81.9 | — | — | — | — | |
| Brain-OF Base2026.02 | 81.88 | — | — | — | — | |
| FEMBA-HugeSize=386M2026.04 | 81.82 | 90.05 | 89.21 | — | — | |
| CSBrainParameters=4.9M, Training Protocol=Single-Task2026.05 | 81.7 | — | — | — | — | |
| CSBrainModel Parameter=4.9M, Task Protocol=Single-Task2026.05 | 81.7 | 90.1 | 89.6 | — | — | |
| CEReBrO-HugeSize=85.15M2026.04 | 81.67 | 90.49 | 89.16 | — | — | |
| LaBraM-MAttBackbone foundation model=LaBraM, Attention mechanism variant=MAtt2026.06 | 81.62 | — | — | — | — | |
| LUNA-HugeSize=311.4M2026.04 | 81.57 | 90.29 | 89.57 | — | — | |
| DARE-EEG-DeepModel variant=Deep2026.05 | 81.56 | 90.6 | 89.04 | — | — | |
| PCA + DSDistillation ratio (%)=252025.12 | 81.5 | 90.3 | 90.4 | — | — | |
| EEG-DLiteDistillation ratio (%)=102025.12 | 81.5 | 90.6 | 90.8 | — | — | |
| LaBraMProtocol=Linear probing2026.05 | 81.5 | — | 87.6 | — | — | |
| DARE-EEG-BaseModel variant=Base2026.05 | 81.45 | 91.38 | 89.27 | — | — | |
| Full dataDistillation ratio (%)=1002025.12 | 81.4 | 89.7 | 90.2 | — | — | |
| LaBraM2026.02 | 81.4 | — | — | — | — | |
| LaBraMPretrain Dataset (Hours)=25002026.03 | 81.4 | — | — | 89.65 | 90.22 | |
| LaBraM-BaseSize=5.9M2026.04 | 81.4 | 89.65 | 90.22 | — | — | |
| LaBraM2025.06 | 81.4 | 89.65 | 90.22 | — | — | |
| LaBraM-BaseParameters=5.8M, Training Protocol=Single-Task2026.05 | 81.4 | — | — | — | — | |
| LaBraM-BaseModel Parameter=5.8M, Task Protocol=Single-Task2026.05 | 81.4 | 89.7 | 90.2 | — | — | |
| LaBraM-BaseBackbone foundation model=LaBraM, Attention mechanism variant=Base2026.06 | 81.4 | — | — | — | — | |
| PCA + DSDistillation ratio (%)=52025.12 | 81.3 | 89.9 | 90.6 | — | — | |
| EEG-DLiteDistillation ratio (%)=252025.12 | 81.3 | 90.1 | 90.2 | — | — | |
| KAST-BAR-BaseParameters=0.8B, Training Protocol=Multi-Task2026.05 | 81.3 | — | — | — | — | |
| KAST-BAR-BaseModel Parameter=0.8B, Task Protocol=Multi-Task2026.05 | 81.3 | 85.6 | 88.2 | — | — | |
| LaBraM# Trainable Params=5.8M, Training Paradigm=Single-task, Finetuning Strategy=Full2026.04 | 81.26 | 89.11 | 88.43 | — | — | |
| PanLUNASize=5.4M, Adaptation Strategy=Full fine-tuning (FF)2026.04 | 81.21 | 89.99 | 89.32 | — | — | |
| RandomDistillation ratio (%)=102025.12 | 81.2 | 90.4 | 90.6 | — | — | |
| MTEEG-DC# Trainable Params=1.1M, Training Paradigm=Multi-task2026.04 | 81.18 | 88.41 | 88.46 | — | — | |
| RandomDistillation ratio (%)=252025.12 | 81.1 | 90.2 | 90.3 | — | — | |
| REVEProtocol=Linear probing2026.05 | 81.1 | — | 90.4 | — | — | |
| LaBraM# Trainable Params=0.3M, Training Paradigm=Single-task, Finetuning Strategy=LoRA2026.04 | 81.05 | 89.44 | 88.24 | — | — | |
| CBraMod-MTDP 100%Pretrain Dataset (Hours)=90002026.03 | 81.02 | — | — | 88.98 | 88.06 | |
| PCA + DSDistillation ratio (%)=102025.12 | 81 | 90.1 | 90.2 | — | — | |
| LUNA-LargeSize=43M2026.04 | 80.96 | 89.86 | 89.24 | — | — | |
| MTEEG-SP# Trainable Params=1.8M, Training Paradigm=Multi-task2026.04 | 80.96 | 87.75 | 87.84 | — | — | |
| RandomDistillation ratio (%)=52025.12 | 80.7 | 89.9 | 90 | — | — | |
| EEG-DLiteDistillation ratio (%)=52025.12 | 80.7 | 89.5 | 90.3 | — | — | |
| LUNA-BaseSize=7M2026.04 | 80.63 | 89.53 | 88.68 | — | — | |
| HPS# Trainable Params=6.0M, Training Paradigm=Multi-task2026.04 | 80.52 | 87.4 | 87.59 | — | — | |
| EEGPT2025.06 | 80.38 | 88.91 | 88.11 | — | — | |
| BIOT-GBWAttBackbone foundation model=BIOT, Attention mechanism variant=GBWAtt2026.06 | 80.16 | — | — | — | — | |
| M3DDistillation ratio (%)=102025.12 | 80.1 | 86.9 | 88.3 | — | — | |
| EEG-DLiteDistillation ratio (%)=12025.12 | 80 | 87.9 | 88.7 | — | — | |
| EEGPTModel variant=large2026.05 | 79.83 | — | 87.18 | — | — | |
| CBraMod-MTDP 25%Pretrain Dataset (Hours)=22502026.03 | 79.81 | — | — | 88.3 | 87.65 | |
| BIOT-MAttBackbone foundation model=BIOT, Attention mechanism variant=MAtt2026.06 | 79.72 | — | — | — | — | |
| NeuroLM-XLParameters=1.7B, Training Protocol=Multi-Task2026.05 | 79.7 | — | — | — | — | |
| NeuroLM-XLModel Parameter=1.7B, Task Protocol=Multi-Task2026.05 | 79.7 | 72.2 | 78.8 | — | — | |
| ST-Transformer2026.02 | 79.66 | — | — | — | — | |
| ST-TransformerPretrain Dataset (Hours)=02026.03 | 79.66 | — | — | 85.21 | 87.07 | |
| ST-Transformer# Trainable Params=3.5M, Training Paradigm=Single-task2026.04 | 79.66 | 85.21 | 87.07 | — | — | |
| ST-Transformer2025.06 | 79.66 | 85.21 | 87.07 | — | — | |
| ST-Trans2026.05 | 79.66 | 85.21 | 87.07 | — | — | |
| MTEEG-RT# Trainable Params=1.8M, Training Paradigm=Multi-task2026.04 | 79.64 | 86.51 | 86.98 | — | — | |
| BIOTParameters=3.2M, Training Protocol=Single-Task2026.05 | 79.6 | — | — | — | — | |
| BIOTModel Parameter=3.2M, Task Protocol=Single-Task2026.05 | 79.6 | 87.9 | 88.2 | — | — | |
| BIOTPretrain Dataset (Hours)=1222026.03 | 79.59 | — | — | 87.92 | 88.15 | |
| BIOTSize=3.2M2026.04 | 79.59 | 86.92 | 88.15 | — | — | |
| BIOT# Trainable Params=3.2M, Training Paradigm=Single-task2026.04 | 79.59 | 87.92 | 88.15 | — | — | |
| BIOT2025.06 | 79.59 | 87.92 | 88.15 | — | — | |
| BIOTPre-training setting=Other2026.05 | 79.59 | 87.92 | 88.15 | — | — | |
| EEGPTModel variant=tiny2026.05 | 79.59 | — | 87.16 | — | — | |
| BIOT-BaseBackbone foundation model=BIOT, Attention mechanism variant=Base2026.06 | 79.59 | — | — | — | — | |
| RandomDistillation ratio (%)=12025.12 | 79.5 | 87.7 | 88.2 | — | — | |
| M3DDistillation ratio (%)=52025.12 | 79.5 | 86.2 | 88.1 | — | — | |
| ST-TransformerParameters=3.5M, Training Protocol=Single-Task2026.05 | 79.3 | — | — | — | — | |
| ST-TransformerModel Parameter=3.5M, Task Protocol=Single-Task2026.05 | 79.3 | 85.4 | 86.9 | — | — | |
| EEGPTParameters=25M, Training Protocol=Multi-Task2026.05 | 79.2 | — | — | — | — | |
| EEGPTModel Parameter=25M, Task Protocol=Multi-Task2026.05 | 79.2 | 74.6 | 86.6 | — | — | |
| BIOTPre-training setting=PREST2026.05 | 79.07 | 87.52 | 87.3 | — | — | |
| SPaRCNetPretrain Dataset (Hours)=02026.03 | 78.96 | — | — | 84.14 | 86.76 | |
| SPaRCNet# Trainable Params=0.79M, Training Paradigm=Single-task2026.04 | 78.96 | 84.14 | 86.76 | — | — | |
| SPaRCNet2026.05 | 78.96 | 84.14 | 86.76 | — | — | |
| CBraMod2026.02 | 78.91 | — | — | — | — | |
| CBraModParameters=4.0M, Training Protocol=Single-Task2026.05 | 78.9 | — | — | — | — | |
| CBraModModel Parameter=4.0M, Task Protocol=Single-Task2026.05 | 78.9 | 86.4 | 86.1 | — | — | |
| BIOTProtocol=Linear probing2026.05 | 78.6 | — | 85.3 | — | — | |
| CBraMod 25%Pretrain Dataset (Hours)=22502026.03 | 78.53 | — | — | 87.38 | 86.87 | |
| FFCL2026.02 | 78.48 | — | — | — | — | |
| FFCLPretrain Dataset (Hours)=02026.03 | 78.48 | — | — | 84.48 | 85.69 | |
| FFCL# Trainable Params=2.4M, Training Paradigm=Single-task2026.04 | 78.48 | 84.48 | 85.69 | — | — | |
| FFCL2026.05 | 78.48 | 84.48 | 85.69 | — | — | |
| BrainHarmonix-F2026.02 | 77.94 | — | — | — | — | |
| PCA + DSDistillation ratio (%)=12025.12 | 77.8 | 85 | 86.2 | — | — | |
| CNN-Transformer2026.02 | 77.77 | — | — | — | — |