Online Transfer Learning in Reinforcement Learning Domains
By Yusen Zhan, Matthew E. Taylor. 2015
34 citations recorded by the source corpus.
This paper proposes an online transfer framework to capture the interaction among agents and shows that current transfer learning in reinforcement learning is a special case of online transfer. Furthermore, this paper re-characterizes existing agents-teaching-agents methods as online transfer and analyze one such teaching method in three ways. First, the convergence of Q-learning and Sarsa with tabular representation with a finite budget is proven. Second, the convergence of Q-learning and Sarsa with linear function approximation is established. Third, the we show the asymptotic performance cannot be hurt through teaching. Additionally, all theoretical results are empirically validated.
Subfields: Artificial Intelligence, Control and Systems Engineering
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Kaixiang Lin
- Andrea Gesmundo
- Catherine Wong
- Anil K. Jain
- Jiayu Zhou
- Yifeng Lu
- Neil Houlsby
- Zhuangdi Zhu
In Control and Systems Engineering
- Catherine Wong
- Anil K. Jain
- Jiayu Zhou
- Yifeng Lu
- Andrea Gesmundo
- Neil Houlsby
- Kaixiang Lin
- Zhuangdi Zhu
Citations in the corpus
- Transfer Learning with Neural AutoML (2018, methodology): “Transfer learning has achieved excellent results as an initialization method for deep networks, including for models trained using RL [24, 25, 26].”
- Transfer Learning in Deep Reinforcement Learning: A Survey (2020, methodology): “g, learning the policy π, the value function V , or even the transition dynamics T (for model-based RL), can benefit from the transferred knowledge? Is the mapping learned offline or online [46]? • What goal to achieve for the transfer learning approach: We can answer this question by analyzing two ”