Research map / Label-efficient and robust learning / Recognition and long-tail
Aligning individual brains with fused unbalanced Gromov Wasserstein
Alexis Thual, Quang Huy TRAN, Tatiana Zemskova, Nicolas Courty, Rémi Flamary, Stanislas Dehaene, Bertrand Thirion
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 |
|---|---|---|---|
| Align subject pairs using FUGW rather than anatomical projection | IBC subjects on fsaverage5 · IBC test contrasts | Pearson correlation: 0.258 to 0.356 | Improved |
| Use FUGW rather than MSM for pairwise alignment | IBC subjects on fsaverage5 · IBC test contrasts | Pearson correlation gain: 0.12 vs 0.01 | Improved |
| Train FUGW with 20 PCA components instead of all ALL-MATH contrasts (ablation) | FUGW · IBC Mathlang test contrasts | Pearson correlation gain: 0.12 -> 0.11 | No clear effect |
| Train FUGW with the smaller ARCHI set instead of ALL-MATH (ablation) | FUGW · IBC Mathlang test contrasts | Pearson correlation gain: 0.12 -> 0.10 | No clear effect |
| Train FUGW with ARCHI PCA instead of ALL-MATH (ablation) | FUGW · IBC Mathlang test contrasts | Pearson correlation gain: 0.12 -> 0.11 | No clear effect |
| Use Gaussian smoothing instead of FUGW alignment (ablation) | Source functional maps · IBC test contrasts | Pearson correlation gain: 0.03 vs 0.12 | Hurt |
| Remove the anatomical geometry term by setting alpha to 0 (ablation) | FUGW · IBC subjects on fsaverage5 | vertex displacement: up to 100mm | Hurt |
| Remove functional information by setting alpha to 1 (ablation) | FUGW · IBC subjects on fsaverage5 | Pearson correlation | No clear effect |
| Fuse functional and geometric costs rather than using either alone (ablation) | FUGW · IBC subjects on fsaverage5 | Pearson correlation | Improved |
| Reduce the unbalanced mass penalty rho (ablation) | FUGW · IBC subjects on fsaverage5 | Pearson correlation | Improved |
| Use FUGW alignment on individual anatomies rather than projecting onto a common template | IBC subject pairs · Individual cortical meshes | Pearson correlation: about 35% gain | Improved |
| Align subjects to a FUGW barycenter rather than use per-vertex group averaging | IBC group maps · IBC test contrasts | number of supra-threshold vertices: p-value = 3\cdot10^{-4} | Improved |
Closest papers by meaning
- MindAligner: Explicit Brain Functional Alignment for Cross-Subject Visual Decoding from Limited fMRI Data (ICML 2025 · similarity 0.85)
- Wills Aligner: Multi-Subject Collaborative Brain Visual Decoding (AAAI 2025 · similarity 0.84)
- SIM: Surface-based fMRI Analysis for Inter-Subject Multimodal Decoding from Movie-Watching Experiments (ICLR 2025 · similarity 0.84)
- Platonic Representations in the Human Brain: Unsupervised Recovery of Universal Geometry (NeurIPS 2026 · similarity 0.84)
- Sparse components distinguish visual pathways & their alignment to neural networks (ICLR 2025 · similarity 0.84)
- FlatClip: Reusing Image Foundation Models for fMRI Representation Learning via Cortical Flatmaps (NeurIPS 2026 · similarity 0.84)
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.
