Research map / Label-efficient and robust learning / Self-supervised and contrastive
Adversarial Feature Learning
Read the paperCheck 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 |
|---|---|---|---|
| Use BiGAN encoder features instead of the strongest listed non-BiGAN alternative | Unsupervised feature learning on permutation-invariant MNIST · Permutation-invariant MNIST test | 1NN classification accuracy (%): 97.39 vs 97.63 | No clear effect |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · ImageNet LSVRC validation, conv1 transferred | Classification accuracy (%): 56.2 -> 55.3 | Hurt |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · ImageNet LSVRC validation, conv2 transferred | Classification accuracy (%): 54.4 -> 53.2 | Hurt |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · ImageNet LSVRC validation, conv3 transferred | Classification accuracy (%): 49.4 -> 49.3 | No clear effect |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · ImageNet LSVRC validation, conv4 transferred | Classification accuracy (%): 43.9 -> 44.4 | Helped |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · ImageNet LSVRC validation, conv5 transferred | Classification accuracy (%): 33.3 -> 34.8 | Helped |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · PASCAL VOC 2007 classification, fc8 | mAP (%): 37.5 -> 41.7 | Helped |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · PASCAL VOC 2007 classification, fc6-8 | mAP (%): 48.7 -> 52.5 | Helped |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · PASCAL VOC 2007 classification, all layers | mAP (%): 58.9 -> 60.3 | Helped |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · PASCAL VOC 2007 FRCN detection | mAP (%): 46.2 -> 46.9 | Helped |
| Use 112×112 encoder inputs instead of 64×64 (ablation) | BiGAN · PASCAL VOC 2012 FCN segmentation | mIU (%): 34.9 -> 35.2 | Helped |
Closest papers by meaning
- Generalized Adversarially Learned Inference (AAAI 2021 · similarity 0.89)
- Complexity Matters: Rethinking the Latent Space for Generative Modeling (NeurIPS 2023 · similarity 0.88)
- Improving Generative Adversarial Networks via Adversarial Learning in Latent Space (NeurIPS 2022 · similarity 0.88)
- Teaching a GAN What Not to Learn (NeurIPS 2020 · similarity 0.88)
- A Unified Generative Adversarial Network Training via Self-Labeling and Self-Attention (ICML 2021 · similarity 0.88)
- Augmentation-Aware Self-Supervision for Data-Efficient GAN Training (NeurIPS 2023 · similarity 0.88)
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.