Research map / Learning theory and algorithms / Sample complexity
Special Unitary Parameterized Estimators of Rotation
Akshay Chandrasekhar
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 2-vec instead of Gram-Schmidt representation | network rotation head · ModelNet10-SO3 (Chair) | mean angular error: 13.606 -> 12.544 | Improved |
| use 2-vec instead of Gram-Schmidt representation | network rotation head · ModelNet10-SO3 (Sofa) | mean angular error: 15.015 -> 15.077 | No clear effect |
| use 2-vec instead of Gram-Schmidt representation | network rotation head · Cambridge Landmarks (King's College) | mean rotation error: 3.298 -> 3.085 | Improved |
| use 2-vec instead of Gram-Schmidt representation | network rotation head · Cambridge Landmarks (Shop Facade) | mean rotation error: 6.559 -> 7.118 | Hurt |
| use QMAlg instead of SVD representation | network rotation head · Cambridge Landmarks (King's College) | mean rotation error: 3.292 -> 2.631 | Improved |
| use QMAlg instead of SVD representation | network rotation head · ModelNet10-SO3 (Toilet) | mean angular error: 6.135 -> 6.079 | No clear effect |
| use QMSVD instead of SVD representation | network rotation head · ModelNet10-SO3 (Sofa) | mean angular error: 14.967 -> 13.683 | Improved |
| use Moebius approximation solver G_M instead of eigen solution G_P | Wahba solver · Wahba n=3, noise 0.1 | median angular error: 7.4868 -> 12.608 | Hurt |
| use SUPER two-point solver instead of QUEST (unweighted) | two-point Wahba solver · operation counts | multiplications: 89 -> 29 | Improved |
| use quaternion or Euler output instead of Gram-Schmidt | direct regression · ModelNet10-SO3 (Chair) | mean angular error: 13.606 -> 23.640 | Hurt |
Closest papers by meaning
- An Analysis of SVD for Deep Rotation Estimation (NeurIPS 2020 · similarity 0.85)
- Learning Unitary Operators with Help From u(n) (AAAI 2017 · similarity 0.85)
- Algebraically Rigorous Quaternion Framework for the Neural Network Pose Estimation Problem (ICCV 2023 · similarity 0.84)
- Learning with 3D rotations, a hitchhiker's guide to SO(3) (ICML 2024 · similarity 0.83)
- Asymptotically Stable Quaternion-valued Hopfield-structured Neural Network with Periodic Projection-based Supervised Learning Rules (NeurIPS 2025 · similarity 0.82)
- 3D Equivariant Pose Regression via Direct Wigner-D Harmonics Prediction (NeurIPS 2024 · similarity 0.81)
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.
