Research map / Label-efficient and robust learning / Recognition and long-tail
Rethinking Multiple-Instance Learning From Feature Space to Probability Space
Zhaolong Du, Shasha Mao, Xuequan Lu, Mengnan Qi, Yimeng Zhang, Jing Gu, Licheng Jiao
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Rethinks deep multiple-instance learning by moving instance modeling from feature space to probability space to avoid representation-learning degradation.
Problem
Deep multiple-instance learning (MIL) methods solve the bag-level multi-instance problem in feature space, which the authors argue risks algorithm degradation during representation learning as data complexity grows.
Approach
Reframes deep MIL from feature space to probability space, shifting how instances are modeled so that bag-level labels are handled through instance probabilities rather than learned features.
Closest papers by meaning
- Robust Self-Supervised Multi-Instance Learning with Structure Awareness (AAAI 2023 · similarity 0.87)
- Are Multiple Instance Learning Algorithms Learnable for Instances? (NeurIPS 2024 · similarity 0.86)
- A Sphere-Description-Based Approach for Multiple-Instance Learning (TPAMI 2016 · similarity 0.86)
- Attention-based Deep Multiple Instance Learning (ICML 2018 · similarity 0.86)
- Sm: enhanced localization in Multiple Instance Learning for medical imaging classification (NeurIPS 2024 · similarity 0.84)
- Multiple Instance Verification (NeurIPS 2026 · similarity 0.83)
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.
