T3S: Improving Multi-Task Reinforcement Learning with Task-Specific Feature Selector and Scheduler
2026-08-31 • Machine Learning
Machine Learning
AI summaryⓘ
The authors address a common problem in multi-task reinforcement learning where one model tries to learn several tasks at once but struggles because it is unclear which parts of the model should be shared between tasks. They propose a new system called T3S that uses special networks to create task-specific features and a scheduler to decide which tasks to focus on based on how well and how quickly the tasks are learned. Their experiments show that this approach improves learning performance on various robot manipulation tasks compared to previous methods.
multi-task reinforcement learninghypernetworksfeature selectortask schedulertask-specific featuresparameter sharingrobot manipulation taskslearning efficiencytask progresssuccess rate
Authors
Yuanqiang Yu, Tianpei Yang, Yongliang Lv, Yan Zheng, Jianye Hao
Abstract
Multi-task reinforcement learning (MTRL) is a technique to train multiple tasks simultaneously, where previous works usually train a single model to solve different tasks by sharing parameters across various tasks. However, these methods are faced with inter-task interference since what parameters should be shared across tasks is not addressed, dramatically reducing learning efficiency. To solve these problems, we propose a novel MTRL framework called Task-Specific feature Selector and Scheduler (T3S), which consists of two components: a feature selector and a task scheduler. Specifically, the feature selectors employ hypernetworks to construct task-specific soft masks, which can be applied by globally shared representation to construct task-specific features. The task scheduler selects tasks for learning through two metrics, where the selection probability is inversely proportional to task progress (e.g., success rate) and task learning speed. Experimental results show that T3S consistently outperforms the state-of-the-art MTRL algorithms on various robotics manipulation tasks.