Cross-domain transfer for reinforcement learning

By Matthew E. Taylor, Peter Stone. 2007

222 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Theory and Mathematics, Reinforcement Learning, Representation & Self-Supervised Learning

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

In Computational Theory and Mathematics

In Reinforcement Learning

In Representation & Self-Supervised Learning

Citations in the corpus

Learning Parameterized Skills (2012, methodology): “Existing work has shown that it is possible to transfer information between pairs of related problems [76, 75, 44, 3] and that parameterized policies can be constructed to deal with slight variations of a domain [59, 53, 33, 27].”
Meta Reinforcement Learning with Latent Variable Gaussian Processes (2018, background): “Much of the literature on meta and transfer learning in RL has focused on multi-task learning, i.e., cases where the system/robot is the same, but the task changes [16, 29, 3, 5, 20, 21, 24, 8, 12].”
Transfer Learning via Inter-Task Mappings for Temporal Difference Learning (2007, methodology): “Knight Joust is a variation on a previously introduced task (Taylor and Stone, 2007) situated in the grid world domain.”
Autonomous transfer for reinforcement learning (2008, background): “While other work has demonstrated that cross-domain transfer is possible [19], it is likely that autonomously learning mappings will become more difficult when the source and target tasks become less similar.”
Multi-task policy search for robotics (2014, methodology): “This approach has been successfully applied in RL [22] and robotics [18].”
Combining manual feedback with subsequent MDP reward signals for reinforcement learning (2010, result): “From a policy πs derived from a learned source task, Taylor and Stone [17] add πs(s) as an extra action (similar to Method 5), increase the Q-value of the Q(s, πs(s)) (similar to Method 4), and add a state value that takes the value πs(s) (similar to Method 2), comparing the results of these three t”
A Survey on Transfer Learning (2010, background): “There has been a large amount of work on transfer learning for reinforcement learning in the machine learning literature (e.g., [9], [10]).”
MT-Opt: Continuous Multi-Task Robotic Reinforcement Learning at Scale (2021, methodology): “Combining multiple task policies has been explored in reinforcement learning by using gating networks [76, 50], conditioning policies on tasks [15], mapping tasks to parameters of a policy [13, 38, 82], distilling separate task policies into a shared multi-task policy [40, 77, 62, 53, 27, 5].”
Learning to Recognize Activities from the Wrong View Point (2008, methodology): “One might: use similar tasks to reasonably estimate the prior [48,30,?]; use hierarchical bayesian models with hyper priors constrained to be similar for similar tasks [5,25,44]; consider a regularized multi task learning framework, when there is a trade-off between large margins and similar decisio”
An Introduction to Intertask Transfer for Reinforcement Learning (2011, methodology): “Many metrics to measure the benefits of transfer in the target task are possible, such as those shown in Figure 4(b) (replicated from our past transfer learning work (Taylor & Stone 2007a)): 1.”
Transfer Learning for Reinforcement Learning Domains: A Survey (2009, methodology): “Similar to strategy transfer, Taylor and Stone (2007b) learnruleswith RIPPER (Cohen, 1995) that summarize a learned source task policy.”
Grounding Language for Transfer in Deep Reinforcement Learning (2017, unknown): “Most previous approaches have either explored skill transfer (Konidaris & Barto, 2007; Konidaris, Scheidwasser, & Barto, 2012) or value function/policy transfer (Liu & Stone, 2006; Taylor, Stone, & Liu, 2007; Taylor & Stone, 2007).”
Integrating reinforcement learning with human demonstrations of varying ability (2011, methodology): “In this section we present HAT , our approach to combining LfD and RL. HAT consists of three steps, motivated by those used in Rule Transfer: Demonstration The agent performs the task under the teleoper-ated control by a human teacher, or by executing an existing suboptimal controller.”
Learning from Demonstration for Shaping through Inverse Reinforcement Learning (2016, background): “introduced Human - Agent Transfer (HAT), a transfer learning algorithm that employs rule transfer [23] in order to summarize a set of human demonstrations for bootstrapping a reinforcement learning agent’s performance.”