Research map / Label-efficient and robust learning / Recognition and long-tail
Mining Multi-Label Samples from Single Positive Labels
Youngin Cho, Daejin Kim, Mohammad Azam Khan, Jaegul Choo
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 |
|---|---|---|---|
| apply latent adaptation (method i) to S2M sampling (ablation) | S2M sampling · CelebA-ABMNSW B+M+S-A-N-W | sampling iterations to converge: 320 -> 20 | Improved |
| apply latent adaptation (method ii, attribute by attribute) (ablation) | S2M sampling · CelebA-ABMNSW B+M+S-A-N-W | sampling iterations to converge: 320 -> 15 | Improved |
| apply latent adaptation (ii) on class B+M+N+S-A-W (ablation) | S2M sampling · CelebA-ABMNSW | sampling iterations to converge: 165 -> 5 | Improved |
| apply latent adaptation (i) on class B+M+N+S-A-W (ablation) | S2M sampling · CelebA-ABMNSW | sampling iterations to converge: 165 -> 35 | Improved |
| apply latent adaptation (ii) on class A+N+W-B-M-S (ablation) | S2M sampling · CelebA-ABMNSW | sampling iterations to converge: 335 -> 5 | Improved |
| apply latent adaptation (i) on class A+N+W-B-M-S (ablation) | S2M sampling · CelebA-ABMNSW | sampling iterations to converge: 335 -> 40 | Improved |
| apply latent adaptation (ii) on class B+M+S-A-N-W (ablation) | S2M sampling · CelebA-ABMNSW | accuracy: 53.51% -> 59.08% | Improved |
| apply latent adaptation (i) on class B+M+N+S-A-W (ablation) | S2M sampling · CelebA-ABMNSW | accuracy: 43.78% -> 42.70% | No clear effect |
| apply latent adaptation (ii) on class B+M+S-A-N-W (ablation) | S2M sampling · CelebA-ABMNSW | FID: 44.72 -> 46.41 | Hurt |
| apply latent adaptation (ii) on class B+M+N+S-A-W (ablation) | S2M sampling · CelebA-ABMNSW | FID: 38.64 -> 37.15 | Improved |
| add S2M sampling to unconditional GAN vs second-best method | GAN · MNIST-Even | accuracy: 87.87 -> 96.37 | Improved |
| add S2M sampling to unconditional GAN vs second-best method | GAN · FMNIST-Even | accuracy: 91.07 -> 97.87 | Improved |
Closest papers by meaning
- Continuous Conditional Generative Adversarial Networks: Novel Empirical Losses and Label Input Mechanisms (TPAMI 2022 · similarity 0.90)
- CcGAN: Continuous Conditional Generative Adversarial Networks for Image Generation (ICLR 2021 · similarity 0.89)
- Fusing Conditional Submodular GAN and Programmatic Weak Supervision (AAAI 2024 · similarity 0.88)
- Semi-Supervised Multi-View Multi-Label Learning with View-Specific Transformer and Enhanced Pseudo-Label (AAAI 2025 · similarity 0.86)
- Turning Waste into Wealth: Leveraging Low-Quality Samples for Enhancing Continuous Conditional Generative Adversarial Networks (AAAI 2024 · similarity 0.86)
- Addressing Multi-Label Learning with Partial Labels: From Sample Selection to Label Selection (AAAI 2025 · 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.
