自适应正则化让模仿学习更灵活，性能接近专家水平。
RIZE: Adaptive Regularization for Imitation Learning
- 用动态调整的目标约束奖励，避免固定结构僵化
- 在复杂环境上达到专家级表现，尤其擅长少样本场景
- 适合研究模仿学习与奖励设计的开发者
我们提出一种新型逆强化学习方法（RIZE），克服了固定奖励结构的僵硬性与隐式正则化的灵活性不足。基于最大熵逆强化学习框架，引入平方时序差分（TD）正则项，其目标值在训练中动态演化，对恢复出的奖励施加自适应约束，提升决策鲁棒性。为捕捉更丰富的回报信息，将分布强化学习融入学习过程。实验表明，该方法在复杂MuJoCo和Adroit环境中实现专家级性能，在Humanoid-v2任务上使用有限专家示范仍优于基线方法。大量实验与消融分析验证了方法有效性，并揭示了模仿学习中的奖励动态特性。代码已开源：https://github.com/adibka/RIZE。
原文摘要 · Abstract (English)
We propose a novel Inverse Reinforcement Learning (IRL) method that mitigates the rigidity of fixed reward structures and the limited flexibility of implicit reward regularization. Building on the Maximum Entropy IRL framework, our approach incorporates a squared temporal-difference (TD) regularizer with adaptive targets that evolve dynamically during training, thereby imposing adaptive bounds on recovered rewards and promoting robust decision-making. To capture richer return information, we integrate distributional RL into the learning process. Empirically, our method achieves expert-level performance on complex MuJoCo and Adroit environments, surpassing baseline methods on the Humanoid-v2 task with limited expert demonstrations. Extensive experiments and ablation studies further validate the effectiveness of the approach and provide insights into reward dynamics in imitation learning. Our source code is available at https://github.com/adibka/RIZE.
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。