用强化学习动态调整多个教师模型的权重，提升学生模型性能。
Multi-Teacher Knowledge Distillation with Reinforcement Learning for Visual Recognition
- 构建包含教师表现和师生差距的状态空间，由智能体决策权重分配。
- 在图像分类、检测和分割任务上均达到当前最优效果。
- 适合需要高效多教师知识蒸馏的视觉识别研究者使用。
多教师知识蒸馏（Multi-teacher KD）从教师集合中迁移多样化知识至学生网络。其核心挑战在于如何平衡不同教师的蒸馏强度。现有方法多基于教师自身表现或师生差异进行加权，缺乏全面信息指导。本文提出基于强化学习的多教师知识蒸馏（MTKD-RL），将教师表现与师生差距作为智能体的状态输入，由智能体输出教师权重，并根据学生反馈的奖励信号进行更新。该框架通过强化学习机制增强师生互动，实现更优匹配与更具意义的权重分配。在图像分类、目标检测和语义分割等视觉识别任务上的实验表明，MTKD-RL优于现有方法，达到当前最优性能。
原文摘要 · Abstract (English)
Multi-teacher Knowledge Distillation (KD) transfers diverse knowledge from a teacher pool to a student network. The core problem of multi-teacher KD is how to balance distillation strengths among various teachers. Most existing methods often develop weighting strategies from an individual perspective of teacher performance or teacher-student gaps, lacking comprehensive information for guidance. This paper proposes Multi-Teacher Knowledge Distillation with Reinforcement Learning (MTKD-RL) to optimize multi-teacher weights. In this framework, we construct both teacher performance and teacher-student gaps as state information to an agent. The agent outputs the teacher weight and can be updated by the return reward from the student. MTKD-RL reinforces the interaction between the student and teacher using an agent in an RL-based decision mechanism, achieving better matching capability with more meaningful weights. Experimental results on visual recognition tasks, including image classification, object detection, and semantic segmentation tasks, demonstrate that MTKD-RL achieves state-of-the-art performance compared to the existing multi-teacher KD works.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。