Research map / Label-efficient and robust learning / Self-supervised and contrastive
Learning Generalized Transformation Equivariant Representations Via AutoEncoding Transformations
Guo-Jun Qi, Liheng Zhang, Feng Lin, Xiao Wang
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 |
|---|---|---|---|
| replace deterministic AET with probabilistic AVT | AET-project · CIFAR-10 conv classifier | error rate: 7.82 -> 7.75 | Improved |
| replace affine with projective transformations | AET · CIFAR-10 conv classifier | error rate: 8.05 -> 7.82 | Improved |
| replace RotNet with AET-project | RotNet + conv · CIFAR-10 | error rate: 8.84 -> 7.82 | Improved |
| replace RotNet with AVT-project | RotNet · CIFAR-10 KNN K=5 | error rate: 25.01 -> 21.62 | Improved |
| use AVT instead of RotNet with few labels (1000 per class) | RotNet + conv · CIFAR-10 few labels | error rate: 13.57 -> 12.27 | Improved |
| use AVT with 20 labels per class vs RotNet | RotNet + conv · CIFAR-10 20 labels/class | error rate: 35.37 -> 35.44 | No clear effect |
| replace RotNet with AVT-project | RotNet · ImageNet Conv5 non-linear | top-1 accuracy: 43.8 -> 48.4 | Improved |
| replace AET with AVT-project | AET-project · ImageNet linear Conv5 | top-1 accuracy: 37.7 -> 39.1 | Improved |
| replace RotNet with AVT-project | RotNet · Places Conv5 | top-1 accuracy: 33.7 -> 35.6 | Improved |
| replace Mean Teacher with SAT | Mean Teacher · CIFAR-10 4000 labels | error rate: 12.31 -> 9.58 | Improved |
| replace Mean Teacher with SAT | Mean Teacher · SVHN 250 labels | error rate: 4.35 -> 4.30 | No clear effect |
| add entropy minimization (ablation) | SAT · CIFAR-10 4000 labels | error rate: 10.90 -> 9.58 | Improved |
Closest papers by meaning
- AVT: Unsupervised Learning of Transformation Equivariant Representations by Autoencoding Variational Transformations (ICCV 2019 · similarity 0.96)
- Co-Attentive Equivariant Neural Networks: Focusing Equivariance On Transformations Co-Occurring in Data (ICLR 2020 · similarity 0.88)
- Self-supervised Transformation Learning for Equivariant Representations (NeurIPS 2024 · similarity 0.87)
- Group Equivariant Subsampling (NeurIPS 2021 · similarity 0.87)
- Neural Isometries: Taming Transformations for Equivariant ML (NeurIPS 2024 · similarity 0.86)
- Interpreting Equivariant Representations (ICML 2024 · 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.
