Research map / Vision-language and multimodal / Human and embodied modeling
Learning Topology-Agnostic EEG Representations with Geometry-Aware Modeling
Ke Yi, Yansen Wang, Kan Ren, Dongsheng Li
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 |
|---|---|---|---|
| add multi-dimensional position encoding (ablation) | MMM · SEED | accuracy (%): 90.96 -> 91.46 | Improved |
| add multi-level channel hierarchy (region tokens) (ablation) | MMM · SEED | accuracy (%): 91.46 -> 95.28 | Improved |
| add multi-stage pre-training (masking strategies) (ablation) | MMM · SEED | accuracy (%): 95.28 -> 95.15 | Hurt |
| add multi-stage pre-training (masking strategies) (ablation) | MMM · SEED-Lite | accuracy (%): 94.29 -> 94.62 | Improved |
| pre-train on larger union dataset with unified topology (ablation) | MMM · SEED | accuracy (%): 94.59 -> 95.76 | Improved |
| pre-train on SEED-Union with MAE (no unified topology) (ablation) | MAE · SEED | accuracy (%): 90.96 (SEED) vs 90.59 (Union) | Mixed |
| pre-train on TUEG then fine-tune on SEED (ablation) | MMM · SEED | accuracy (%): 94.59 -> 95.29 | Improved |
| pre-train on SEED-Lite series then partially fine-tune on 62-channel SEED (ablation) | MMM · SEED partial fine-tune | accuracy (%): 92.54 -> 92.81 (Union-Lite) | Improved |
| pre-train MAE on Lite data then partially fine-tune (ablation) | MAE · SEED partial fine-tune | accuracy (%): 79.82 -> 75.05 | Hurt |
| use 17 spatial regions instead of 1 region (ablation) | MMM · SEED | accuracy (%): 93.95 -> 94.61 | Improved |
| use 17 regions instead of 62 regions (ablation) | MMM · SEED | accuracy (%): 94.53 -> 94.61 | No clear effect |
| replace MAE with MMM | pretraining framework · SEED-IV | accuracy (%): 81.09 -> 87.92 | Improved |
Closest papers by meaning
- THD-BAR: Topology Hierarchical Derived Brain Autoregressive Modeling for EEG Generic Representations (NeurIPS 2025 · similarity 0.87)
- Multi-dataset Joint Pre-training of Emotional EEG Enables Generalizable Affective Computing (NeurIPS 2025 · similarity 0.87)
- LUNA: Efficient and Topology-Agnostic Foundation Model for EEG Signal Analysis (NeurIPS 2025 · similarity 0.86)
- What Does the Brain See? Multiview Neural Representation to Demystify the Brain-Visual Alignment (NeurIPS 2026 · similarity 0.85)
- Pre-Training Graph Contrastive Masked Autoencoders are Strong Distillers for EEG (ICML 2025 · similarity 0.85)
- EEGPT: Pretrained Transformer for Universal and Reliable Representation of EEG Signals (NeurIPS 2024 · similarity 0.85)
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.
