Research map / Architectures and efficiency / Pruning and efficient tuning
Riemannian High-Order Pooling for Brain Foundation Models
Chen Hu, Ziheng Chen, Rui Wang, Yefeng Zheng, Nicu Sebe
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 RHOP head | BIOT (from scratch) · TUEV | balanced accuracy: 0.4682 -> 0.5355 | Improved |
| add RHOP head | BIOT (from scratch) · TUAB | AUROC: 0.8691 -> 0.8765 | Improved |
| add RHOP head | BIOT (pretrained) · TUAB | balanced accuracy: 0.7959 -> 0.8102 | Improved |
| add RHOP head | LaBraM-Base fine-tuned · TUAB | balanced accuracy: 0.8140 -> 0.8244 | Improved |
| add RHOP head | LaBraM-Base fine-tuned · TUEV | balanced accuracy: 0.6409 -> 0.6380 (kappa 0.6637 -> 0.6785, F1 0.8312 -> 0.8420) | Mixed |
| add RHOP head to frozen backbone | LaBraM-Base linear head · PhysioP300 | balanced accuracy: 0.6327 -> 0.6517 | Improved |
| add RHOP head to frozen backbone | LaBraM-Base linear head · BCIC2B | balanced accuracy: 0.6840 -> 0.6901 | Improved |
| add iSQRT-COV GCP head instead of RHOP | BIOT (from scratch) · TUEV | balanced accuracy: 0.4682 -> 0.4480 | Hurt |
| add SVD-Padé GCP head | LaBraM-Base · TUEV | balanced accuracy: 0.6409 -> 0.5605 | Hurt |
| add quotient Gaussian embedding (ablation) | LaBraM-Base · TUAB | AUROC: 0.9022 -> 0.9048 | Improved |
| add Riemannian Gaussian embedding (ablation) | RHOP (QGE) · TUAB | AUROC: 0.9048 -> 0.9069 | Improved |
| add SICE normalization (ablation) | RHOP · TUAB | AUROC: 0.9069 -> 0.9088 | Improved |
Closest papers by meaning
- LAtte: Hyperbolic Lorentz Attention for Joint-Subject EEG Classification (NeurIPS 2026 · similarity 0.88)
- B[FM]$^2$: Brain Foundation Model via Flow Matching with SplitUNet (NeurIPS 2026 · similarity 0.87)
- Riemannian Flow Matching for Brain Connectivity Matrices via Pullback Geometry (NeurIPS 2025 · similarity 0.87)
- CSBrain: A Cross-scale Spatiotemporal Brain Foundation Model for EEG Decoding (NeurIPS 2025 · similarity 0.87)
- BrainOmni: A Brain Foundation Model for Unified EEG and MEG Signals (NeurIPS 2025 · similarity 0.87)
- What do EEG Foundation Models Capture from Human Brain Signals? (NeurIPS 2026 · similarity 0.87)
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.
