Behavior transfer for value-function-based reinforcement learning

By Matthew E. Taylor, Peter Stone. 2005

155 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Management Information Systems

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 Management Information Systems

Citations in the corpus

Transfer via inter-task mappings in policy search reinforcement learning (2007, methodology): “These mappings are analogous to those previously used in Keepaway [13] and are used to construct ρH .”
Keepaway Soccer: From Machine Learning Testbed to Benchmark (2005, methodology): “In addition to our own previous work [16, 18], work by DiPietro et al.”
Curriculum Learning for Reinforcement Learning Domains: A Framework and Survey (2020, methodology): “This knowledge can take the form of samples (Lazaric et al., 2008; Lazaric and Restelli, 2011), options (Soni and Singh, 2006), policies (Fernández et al., 2010), models (Fachantidis et al., 2013), or value functions (Taylor and Stone, 2005).”
Value Functions for RL-Based Behavior Transfer: A Comparative Study (2005, background): “In this paper we study the effect of behavior transfer ( Taylor & Stone 2005 ) on the learning rates of value-based TD learners.”
Policy Transfer using Reward Shaping (2015, background): “static potential-based reward shaping is shown to be equivalent, given the same experiences, to Q-value initialization [24], which is exactly what many transfer learning techniques do [19, 21].”
Batch reinforcement learning in a complex domain (2007, background): “Problems related to reinforcement learning, such as behavior transfer [15] have been studied in the context of Keepaway.”
Reinforcement Learning for RoboCup Soccer Keepaway (2005, methodology): “Using real robots, Uchibe (1999) used reinforcement learning methods to learn to shoot a ball into a goal while avoiding an opponent.”
Autonomous Task Sequencing for Customized Curriculum Design in Reinforcement Learning (2017, background): “A variety of transfer learning methods have been proposed, enabling RL agents to transfer samples [Lazaric et al., 2008], options [Soni and Singh, 2006; Perkins et al., 1999], policies [Fernández et al., 2010], models [Fachantidis et al., 2013], and value functions [Taylor and Stone, 2005].”
An Experts Algorithm for Transfer Learning (2007, methodology): “Following [Taylor and Stone, 2005] we used SARSA(0) with 1D tile-coding and a linear function approximator to train agents in 3v2 keepaway for 2000 episodes and then asked if we could use the resulting policies to do well in 4v2 keepaway.”
Probabilistic policy reuse in a reinforcement learning agent (2006, background): “Transfer learning in more complex domains, as the Keepaway task in robot soccer, requires a mapping between tasks that use different state and action spaces [15, 14].”
Transfer in Reinforcement Learning: A Framework and a Survey (2012, background): “The area ratio metric introduced by Taylor and Stone (2009) copes with this problem by considering the whole area under the learning curves with and without transfer.”
General Game Learning Using Knowledge Transfer (2007, background): “Knowledge transfer has received significant attention recently in machine learning research [Asgharbeygi et al., 2006; Taylor and Stone, 2005; Ferns et al., 2006].”
Transfer Learning via Inter-Task Mappings for Temporal Difference Learning (2007, background): “This article contains and extends material published in two conference papers (Taylor and Stone, 2005; Taylor et al., 2005).”
Source Task Creation for Curriculum Learning (2016, background): “This knowledge can take the form of samples [11] [12], options [21], policies [6], models [5], or value functions [26].”
Multiagent Transfer Learning via Assignment-Based Decomposition (2009, methodology): “More importantly, it enables transfer between domains of different sizes simply by appropriately initializing the value function of one domain from the value function of a previously learned domain in a manner similar to the behavior transfer of [ 6 ].”