Research map / Label-efficient and robust learning / Recognition and long-tail
FiGuRO: Intrinsic Dimension Estimation for Multi-Modal Data
Viktoria Schuster, Sana Tonekaboni, Caroline Uhler
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 R(D) fidelity-guided bidirectional rank adjustment to SVD rank reduction (ablation) | FiGuRO · Simulated B subsets | shared rank k_s (Bs): 1.0 -> 3.6 ± 0.06 | Improved |
| add R(D) to SVD rank reduction (ablation) | FiGuRO · Simulated Bi1 | shared rank k_s (GT 20): 1.0 -> 13.4 ± 0.4 | Improved |
| add R(D) to SVD rank reduction (ablation) | FiGuRO · Simulated Bl | shared rank k_s (GT 20): 1.0 -> 19.2 ± 0.7 | Improved |
| use FiGuRO instead of JIVE/AJIVE/SLIDE/ShIndICA | FiGuRO · Simulated B | avg deviation from GT rank: 9.5 -> 2.9 | Improved |
| add Frobenius norm or L1 regularizer for disentanglement (ablation) | FiGuRO · Simulated B | disentanglement / rank accuracy | Hurt |
| add L2 regularization (ablation) | FiGuRO · Simulated B (fully independent modalities) | disentanglement | Mixed |
| use FiGuRO shared subspace instead of unimodal embedding | FiGuRO · So2Sat | Acc_all: 0.41 -> 0.55 | Improved |
| use FiGuRO instead of AJIVE | FiGuRO · Audio MNIST | Acc_s: 0.99 -> 0.94 | Hurt |
| use FiGuRO instead of SLIDE | FiGuRO · So2Sat | Acc_s: 0.24 -> 0.55 | Improved |
| vary distortion budget lambda, tau, gamma, patience (grid search) (ablation) | FiGuRO · Dataset A (N=1080 runs) | estimated ID (true 5): 4.89 ± 0.01 | No clear effect |
| use R^2 distortion metric instead of MSE/RMSE (ablation) | FiGuRO · Simulated sparse data | robustness | Improved |
| use R^2 distortion metric instead of Explained Variance Score (ablation) | FiGuRO · Simulated nonlinear data | robustness | Improved |
Closest papers by meaning
- Intrinsic Dimension Correlation: uncovering nonlinear connections in multimodal representations (ICLR 2025 · similarity 0.88)
- An Information Criterion for Controlled Disentanglement of Multimodal Data (ICLR 2025 · similarity 0.86)
- RePercENT: Scaling Disentangled Representation Learning Beyond Two Modalities (NeurIPS 2026 · similarity 0.86)
- Multi-View Causal Representation Learning with Partial Observability (ICLR 2024 · similarity 0.86)
- Synergy Between Sufficient Changes and Sparse Mixing Procedure for Disentangled Representation Learning (ICLR 2025 · similarity 0.85)
- Multi-modal contrastive learning adapts to intrinsic dimensions of shared latent variables (NeurIPS 2025 · 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.
