Research map / Label-efficient and robust learning / Recognition and long-tail
Attention-based Deep Multiple Instance 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 trainable attention pooling instead of common MIL methods | Deep MIL · Musk1, Musk2, Fox, Tiger, Elephant | classification accuracy | Mixed |
| Use attention pooling rather than embedding max pooling | LeNet5 MIL · MNIST-bags, mean bag size 10, 50 training bags | AUC: 0.713 -> 0.768 | Helped |
| Use attention pooling rather than embedding max pooling | LeNet5 MIL · MNIST-bags, mean bag size 50, 50 training bags | AUC: 0.872 -> 0.967 | Helped |
| Use gated attention pooling rather than embedding max pooling | LeNet5 MIL · MNIST-bags, mean bag size 100, 50 training bags | AUC: 0.977 -> 0.998 | Helped |
| Replace max pooling with mean pooling (ablation) | Embedding-based MIL · Breast Cancer | accuracy: 0.607 -> 0.741 | Helped |
| Replace max pooling with mean pooling (ablation) | Instance-based MIL · Breast Cancer | AUC: 0.612 -> 0.719 | Helped |
| Replace max pooling with mean pooling (ablation) | Embedding-based MIL · Colon Cancer | AUC: 0.918 -> 0.940 | Helped |
| Replace max pooling with mean pooling (ablation) | Instance-based MIL · Colon Cancer | accuracy: 0.842 -> 0.772 | Hurt |
| Add gated nonlinearity to attention pooling (ablation) | Deep MIL · Breast Cancer | accuracy: 0.745 -> 0.755 | Helped |
| Add gated nonlinearity to attention pooling (ablation) | Deep MIL · Colon Cancer | AUC: 0.968 -> 0.968 | No clear effect |
| Use embedding-based rather than instance-based learning (ablation) | Deep MIL · MNIST-bags | AUC | Helped |
| Increase training data to sufficient numbers of bags and instances | Deep MIL on MNIST-bags · MNIST-bags | AUC | No clear effect |
Closest papers by meaning
- Loss-Based Attention for Deep Multiple Instance Learning (AAAI 2020 · similarity 0.90)
- Are Multiple Instance Learning Algorithms Learnable for Instances? (NeurIPS 2024 · similarity 0.87)
- Robust Self-Supervised Multi-Instance Learning with Structure Awareness (AAAI 2023 · similarity 0.87)
- Mixture of Mini Experts: Overcoming the Linear Layer Bottleneck in Multiple Instance Learning (ICLR 2026 · similarity 0.87)
- Sm: enhanced localization in Multiple Instance Learning for medical imaging classification (NeurIPS 2024 · similarity 0.85)
- How Effective Can Dropout Be in Multiple Instance Learning ? (ICML 2025 · similarity 0.84)
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.