Research map / Architectures and efficiency / Attention and transformers
Scaling Vision Transformers for Functional MRI with Flat Maps
Connor Lane, Mihir Tripathy, Leema Krishna Murali, Ratna Sagari Grandhi, Shamus Sim Zi Yang, Sam Gijsen, Debojyoti Das, Manish Ram, Utkarsh Kumar Singh, Cesar Kadir Torrico Villanueva, YUXIANG WEI, Will Beddow and 6 more
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 |
|---|---|---|---|
| use flat map representation instead of parcellation | CortexMAE · HCP-A Sex | probe accuracy: 71.2 -> 87.4 | Improved |
| use flat map representation instead of parcellation | CortexMAE · HCP-YA Task21 | probe accuracy: 97.5 -> 98.9 | Improved |
| use flat map representation instead of parcellation | CortexMAE · NSD COCO24 | probe accuracy: 27.5 -> 31.0 | Improved |
| use volume representation instead of flat map | CortexMAE · HCP-A Age | probe accuracy: 47.5 -> 53.4 | Improved |
| use flat map representation instead of parcellation | CortexMAE · ABIDE, ADHD200, ADNI, PPMI | probe accuracy: 61.4 vs 62.0 on ABIDE | No clear effect |
| add tube masking instead of uniform masking (ablation) | CortexMAE flat · NSD COCO24 (ratio 90) | accuracy: 29.8 -> 31.4 | Improved |
| add TR scaling augmentation (ablation) | CortexMAE flat · NSD COCO24 | accuracy: 30.4 -> 31.7 | No clear effect |
| add strong random crop augmentation (ablation) | CortexMAE flat · NSD COCO24 | accuracy: 30.4 -> 24.4 | Hurt |
| use PC norm with 2 components as target (ablation) | CortexMAE flat · NSD COCO24 | accuracy: 29.5 -> 32.1 | No clear effect |
| use PC norm with 8 components as target (ablation) | CortexMAE flat · NSD COCO24 | accuracy: 29.5 -> 24.6 | Hurt |
| reduce temporal patch size from 4 to 2 (ablation) | CortexMAE flat · NSD COCO24 | accuracy: 29.6 -> 32.9 | Improved |
| add subcortical structures to parcellation (Schaefer-400 + Tian S3) (ablation) | CortexMAE parcel · NSD COCO24 | accuracy: 27.5 -> 27.2 | No clear effect |
Closest papers by meaning
- FlatClip: Reusing Image Foundation Models for fMRI Representation Learning via Cortical Flatmaps (NeurIPS 2026 · similarity 0.91)
- Meta-Learning an In-Context Transformer Model of Human Higher Visual Cortex (NeurIPS 2025 · similarity 0.88)
- Multimodal Scaling Laws for Task & Data-Optimized Models of Visual Cortex (ICML 2026 · similarity 0.88)
- MindEye2: Shared-Subject Models Enable fMRI-To-Image With 1 Hour of Data (ICML 2024 · similarity 0.87)
- In Silico Mapping of Visual Categorical Selectivity Across the Whole Brain (NeurIPS 2025 · similarity 0.86)
- BrainWorld: A Structural-Prior-Conditioned Generative Model for Whole-Brain 4D fMRI Dynamics (NeurIPS 2026 · 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.
