Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Equivariant Covariance Tensors: Guaranteed SPD Uncertainty for Tensor-Valued Geometric Learning
Ruihan Liu, Yu Ji, Jianbo Yu, Shifu Yan, Qingchao Jiang
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 full covariance instead of diagonal covariance (ablation) | equivariant UQ model · ModelNet40 inertia | MAE: 0.092 -> 0.078 | Improved |
| use full covariance instead of diagonal UQ (ablation) | equivariant UQ model · Materials Project dielectric | MAE: 2.25 -> 1.55 | Improved |
| train with LE-ESO (Laplace) instead of Gaussian NLL (ablation) | matrix-exp covariance head · Materials Project dielectric | MAE: 1.78 -> 1.55 | Improved |
| train with LE-ESO instead of Energy Score (ablation) | matrix-exp covariance head · Materials Project dielectric | MAE: 1.64 -> 1.55 | Improved |
| train with LE-ESO instead of Gaussian NLL (ablation) | matrix-exp covariance head · Materials Project dielectric | MACE: 0.092 -> 0.049 | Improved |
| use matrix exponential head instead of Cholesky covariance head (ablation) | equivariant mean model · ModelNet40 inertia | covariance equivariance error: 4.30e-1 -> 2.75e-7 | Improved |
| use matrix exponential instead of direct equivariant regression of covariance (ablation) | equivariant covariance head · ModelNet40 inertia | SPD validity | Improved |
| apply post-hoc temperature scaling (T about 0.05) (ablation) | Ours (Full Cov.) · Materials Project dielectric | Energy Score: 0.87 -> 0.66 | Improved |
| rank by lambda_max(Sigma) instead of Trace(Sigma) for risk-coverage (ablation) | predicted covariance · Materials Project, 90% coverage | retained-set MAE improvement: 0.8% -> 3.1% | Improved |
| joint training without gradient detachment instead of detached UQ gradients (ablation) | LE-ESO training · Materials Project dielectric | MAE: MAE difference <0.02 | No clear effect |
| set LE-ESO weight alpha=1 versus smaller values (ablation) | LE-ESO · validation sweep | validation MAE | Improved |
| add anisotropic jitter to eigenvalues for gradient stability (ablation) | spectral decomposition · Materials Project dielectric | calibration (MACE): MACE change <1e-4 | No clear effect |
Closest papers by meaning
- Beyond Canonicalization: How Tensorial Messages Improve Equivariant Message Passing (ICLR 2025 · similarity 0.87)
- Permutation Equivariant Neural Networks for Symmetric Tensors (ICML 2025 · similarity 0.87)
- The Price of Freedom: Exploring Expressivity and Runtime Tradeoffs in Equivariant Tensor Products (ICML 2025 · similarity 0.86)
- Intrinsic Riemannian Cross-covariance for Manifold-valued Random Objects (NeurIPS 2026 · similarity 0.85)
- A Tale of Two Symmetries: Exploring the Loss Landscape of Equivariant Models (NeurIPS 2025 · similarity 0.85)
- Learning symmetries via weight-sharing with doubly stochastic tensors (NeurIPS 2024 · 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.
