Research map / Architectures and efficiency / Mixture of experts
NeurIPS: Neuro-anatomical Inductive Priors for Sphere-based Brain Decoding
Sijin Yu, Zijiao Chen, Zhenyu Yang, Zihao Tan, Jiakun Xu, Zhongliang Liu, shengxian chen, WENXUAN WU, Xiangmin Xu, Xin Zhang
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 global token (ablation) | NeurIPS · NSD test ablation | CLIP: 89.4% -> 93.2% | Improved |
| route SG-MoE by anatomy instead of subject ID (ablation) | NeurIPS · NSD test ablation | CLIP: 92.7% -> 93.2% | Improved |
| route by correct anatomy instead of swapped anatomy (ablation) | NeurIPS · NSD test ablation | CLIP: 91.9% -> 93.2% | Improved |
| route by correct anatomy instead of random anatomy (ablation) | NeurIPS · NSD test ablation | CLIP: 92.0% -> 93.2% | Improved |
| add anatomical input to router (ablation) | NeurIPS · NSD test ablation | CLIP: 90.2% -> 93.2% | Improved |
| gate on anatomy instead of functional features (ablation) | NeurIPS · NSD test ablation | CLIP: 92.7% -> 93.2% | Improved |
| restrict tokenizer to visual ROIs instead of full cortex (ablation) | NeurIPS · NSD test ablation | CLIP: 91.0% -> 93.2% | Improved |
| restrict tokenizer to visual ROIs instead of full cortex (ablation) | NeurIPS · NSD test ablation | memory (GB): 74GB -> 61GB | Improved |
| preserve spherical topology (no shuffling) (ablation) | NeurIPS · NSD test ablation | CLIP: 91.5% -> 93.2% | Improved |
| use convolution receptive field beyond 1 (ablation) | NeurIPS · NSD test ablation | CLIP: 91.7% -> 93.2% | Improved |
| add semantic decoder (ablation) | NeurIPS · NSD test ablation | CLIP: 55.0% -> 93.2% | Improved |
| add perception decoder (ablation) | NeurIPS · NSD test ablation | PixCor: 0.148 -> 0.248 | Improved |
Closest papers by meaning
- NeurIPT: Foundation Model for Neural Interfaces (NeurIPS 2025 · similarity 0.87)
- Neuro-Vision to Language: Enhancing Brain Recording-based Visual Reconstruction and Language Interaction (NeurIPS 2024 · similarity 0.87)
- FlatClip: Reusing Image Foundation Models for fMRI Representation Learning via Cortical Flatmaps (NeurIPS 2026 · similarity 0.87)
- MindBridge: A Cross-Subject Brain Decoding Framework (CVPR 2024 · similarity 0.86)
- $i$MIND: Insightful Multi-subject Invariant Neural Decoding (NeurIPS 2025 · similarity 0.86)
- SPINT: Spatial Permutation-Invariant Neural Transformer for Consistent Intracortical Motor Decoding (NeurIPS 2025 · similarity 0.86)
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.
