Research map / Label-efficient and robust learning / Recognition and long-tail
AET vs. AED: Unsupervised Representation Learning by Auto-Encoding Transformations Rather Than Data
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Predict projective transformations from encoded features rather than classify rotations | NIN representations · CIFAR-10 | error rate (conv): 8.84 -> 7.82 | Helped |
| Use projective instead of affine transformations (ablation) | AET · CIFAR-10 | error rate (conv): 8.05 -> 7.82 | Helped |
| Use projective instead of affine transformations (ablation) | AET · CIFAR-10 | error rate (KNN): 23.07 -> 22.39 | Helped |
| Predict projective transformations from encoded features rather than classify rotations | AlexNet representations · ImageNet | Top-1 accuracy (Conv4, nonlinear): 50.0 -> 53.2 | Helped |
| Predict projective transformations from encoded features rather than classify rotations | AlexNet representations · ImageNet | Top-1 accuracy (Conv5, nonlinear): 43.8 -> 47.0 | Helped |
| Pretrain AET features on ImageNet and transfer them to Places | AlexNet representations · Places Conv1 | Top-1 accuracy: 23.3 -> 22.1 | Hurt |
| Pretrain AET features on ImageNet and transfer them to Places | AlexNet representations · Places Conv2 | Top-1 accuracy: 33.9 -> 32.9 | Hurt |
| Pretrain AET features on ImageNet and transfer them to Places | AlexNet representations · Places Conv3 | Top-1 accuracy: 36.3 -> 37.1 | Helped |
| Pretrain AET features on ImageNet and transfer them to Places | AlexNet representations · Places Conv4 | Top-1 accuracy: 34.7 -> 36.2 | Helped |
| Pretrain AET features on ImageNet and transfer them to Places | AlexNet representations · Places Conv5 | Top-1 accuracy: 33.7 -> 34.7 | Helped |
Closest papers by meaning
- AE2-Nets: Autoencoder in Autoencoder Networks (CVPR 2019 · similarity 0.87)
- Semantic-Aware Auto-Encoders for Self-Supervised Representation Learning (CVPR 2022 · similarity 0.86)
- Efficient Self-supervised Vision Transformers for Representation Learning (ICLR 2022 · similarity 0.86)
- Unsupervised Representation Learning by Predicting Image Rotations (ICLR 2018 · similarity 0.86)
- URLOST: Unsupervised Representation Learning without Stationarity or Topology (ICLR 2025 · similarity 0.85)
- Augmenting Supervised Neural Networks with Unsupervised Objectives for Large-scale Image Classification (ICML 2016 · 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.