Research map / Graph learning / GNN architectures
Lightweight Transformer for EEG Classification via Balanced Signed Graph Algorithm Unrolling
Junyi Yao, Parham Eftekhar, Gene Cheung, Xujin Chris Liu, Yao Wang, Wei Hu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| enforce graph balance with signed edges (vs positive graph) (ablation) | unrolled graph denoiser · Turkish Epilepsy EEG (LOSO) | accuracy (%): 84.30 -> 93.68 | Improved |
| enforce graph balance (vs unbalanced signed graph) (ablation) | unrolled graph denoiser · Turkish Epilepsy EEG (LOSO) | accuracy (%): 78.87 -> 93.68 | Improved |
| add contrastive term to MSE denoiser loss (ablation) | unrolled graph denoiser · Turkish Epilepsy EEG (default split) | accuracy (%): 81.44 -> 97.57 | Improved |
| add contrastive term to MSE denoiser loss (ablation) | unrolled graph denoiser · Turkish Epilepsy EEG (default split) | recall / specificity: recall 97.67 -> 95.98; specificity 99.32 -> 97.45 | Hurt |
| use Mahalanobis instead of Euclidean feature distance (ablation) | unrolled graph denoiser · Turkish Epilepsy EEG (default split) | accuracy (%): 96.80 -> 97.57 (convergence 1h55 -> 2h14) | Improved |
| increase temporal sequence length beyond 10 (ablation) | unrolled graph denoiser · Turkish Epilepsy EEG (default split) | accuracy (%): 97.57 (10) -> 88.10 (100) | Hurt |
| increase CNN blocks from 1 to 6 (ablation) | unrolled graph denoiser · Turkish Epilepsy EEG (default split) | accuracy (%): 65.15 -> 97.57 | Improved |
| increase CNN blocks beyond 6 (ablation) | unrolled graph denoiser · Turkish Epilepsy EEG (default split) | accuracy (%): 97.57 -> 97.44 (9), 97.42 (12) | No clear effect |
| use the small unrolled model instead of large STFT + CNN | unrolled graph denoiser · Turkish Epilepsy EEG (default split) | accuracy (%): 97.57 vs 99.20 with 14,787 vs 11,533,928 params | Mixed |
Closest papers by meaning
- REST: Efficient and Accelerated EEG Seizure Analysis through Residual State Updates (ICML 2024 · similarity 0.84)
- Pre-Training Graph Contrastive Masked Autoencoders are Strong Distillers for EEG (ICML 2025 · similarity 0.84)
- EvoBrain: Dynamic Multi-Channel EEG Graph Modeling for Time-Evolving Brain Networks (NeurIPS 2025 · similarity 0.83)
- EEGPT: Pretrained Transformer for Universal and Reliable Representation of EEG Signals (NeurIPS 2024 · similarity 0.83)
- Decoding Natural Images from EEG for Object Recognition (ICLR 2024 · similarity 0.82)
- B[FM]$^2$: Brain Foundation Model via Flow Matching with SplitUNet (NeurIPS 2026 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
