Research map / Label-efficient and robust learning / Recognition and long-tail
Disambiguated Attention Embedding for Multi-Instance Partial-Label Learning
Wei Tang, Weijia Zhang, Min-Ling Zhang
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 |
|---|---|---|---|
| add attention loss (ablation) | DEAM · FMNIST-MIPL, r=1 | accuracy: 0.744 -> 0.881 | Improved |
| add attention loss (ablation) | DEAM · FMNIST-MIPL, r=3 | accuracy: 0.586 -> 0.657 | Improved |
| add attention loss (ablation) | DEAM · SIVAL-MIPL, r=1 | accuracy: 0.607 -> 0.635 | Improved |
| add attention loss (ablation) | DEAM · SIVAL-MIPL, r=3 | accuracy: 0.499 -> 0.503 | No clear effect |
| momentum-based disambiguation vs lambda=0 (progressive, current classifier output) variant (ablation) | DEAM · MNIST/FMNIST/Birdsong-MIPL, larger r | accuracy | Improved |
| momentum-based disambiguation vs lambda=1 (uniform weights) variant (ablation) | DEAM · MNIST/FMNIST/Birdsong-MIPL, larger r | accuracy | Improved |
| uniform weights (lambda=1) instead of progressive (lambda=0) at r=3 (ablation) | DEAM variants · MNIST-MIPL and FMNIST-MIPL | accuracy | Mixed |
| vary attention-loss weight lambda_a (ablation) | DEAM · MNIST-MIPL and CRC-MIPL | accuracy | No clear effect |
| use DEAM instead of MIPLGP on benchmark datasets | MIPL algorithm · MNIST/FMNIST/Birdsong-MIPL, r=1 | accuracy: 0.949 -> 0.976 | Improved |
| use DEAM instead of MIPLGP on SIVAL-MIPL | MIPL algorithm · SIVAL-MIPL, r=1 | accuracy: 0.669 -> 0.635 | Hurt |
| use DEAM instead of MIPLGP on larger r | MIPL algorithm · MNIST-MIPL, r=2 | accuracy: 0.817 -> 0.943 | Improved |
| use DEAM instead of PLL algorithm AGGD (MaxMin) on CRC-MIPL-SIFT | MIPL algorithm · CRC-MIPL-SIFT | accuracy: 0.285 -> 0.532 | Improved |
Closest papers by meaning
- Multi-Instance Partial-Label Learning with Margin Adjustment (NeurIPS 2024 · similarity 0.91)
- Fast Multi-Instance Partial-Label Learning (AAAI 2025 · similarity 0.90)
- Navigating Real-World Partial Label Learning: Unveiling Fine-Grained Images with Attributes (AAAI 2024 · similarity 0.87)
- Distilling Reliable Knowledge for Instance-Dependent Partial Label Learning (AAAI 2024 · similarity 0.87)
- Self-Paced Unified Representation Learning for Hierarchical Multi-Label Classification (AAAI 2024 · similarity 0.86)
- Disentangled Partial Label Learning (AAAI 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.
