Giving Advice about Preferred Actions to Reinforcement Learners Via Knowledge-Based Kernel Regression

By Richard Maclin, Jude W. Shavlik, Lisa A. Torrey, Trevor Walker, Edward W. Wild. 2005

107 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Interactive Learning from Policy-Dependent Human Feedback (2017, background): “Some research has also examined combining human feed-back with more traditional environmental rewards (Knox & Stone, 2010; Tenorio-Gonzalez et al., 2010; Clouse & Utgoff, 1992; Maclin et al., 2005).”
A Survey of Preference-Based Reinforcement Learning Methods (2017, background): “The advice can also be defined via other means, such as rules (Maclin et al., 2005).”
Reinforcement Learning Via Practice and Critique Advice (2010, unknown): “For example , rules in the form of programming-language constructs (Maclin and Shavlik 1996), logical rules derived from a constrained natural language (Kuhlmann et al. 2004), and rules about action utilities and preferences (Maclin et al. 2005b; 2005a).”
Learning From Explanations Using Sentiment and Advice in RL (2017, background): “Various forms of advice have been developed in other work, including linking one condition to each action [12], and linking a condition to rewards [11].”
A comprehensive survey on safe reinforcement learning (2015, methodology): “The approaches in Section 4.1.3 combine the advice provided by the teacher with these exploration strategies to produce a safe exploration process (Clouse, 1997; Maclin et al., 2005a; Thomaz and Breazeal, 2006, 2008; Torrey and Taylor, 2012).”
Transfer Learning for Reinforcement Learning Domains: A Survey (2009, methodology): “In this work, Q-values re learned via support vector regression, and thenPreference Knowledge Based Kernel Regression(KBKR) (Maclin et al., 2005) adds the advice as soft constraints in the target, setting relative preferences for different actions in different states.”
Teachable robots: Understanding human teaching behavior to build more effective robot learners (2008, methodology): “Conceptually, our modiﬁed version gives the algorithm a pre-action and post-action phase in order to incorporate the new guidance input.”
Half Field Offense in RoboCup Soccer: A Multiagent Reinforcement Learning Case Study (2006, background): “Several researchers have focused on specific subtasks like goal-shooting [3, 6].”
Autonomous transfer for reinforcement learning (2008, background): “Any learning method can be used that is capable of utilizing inter-task mappings for transfer (e.g., KBKR [8], Sarsa [12, 14], or NEAT [16]).”
Imitation Learning with Demonstrations and Shaping Rewards (2014, result): “Similar comments hold for other work on advice in RL (Rosenstein and Barto 2004; Maclin et al. 2005).”
Probabilistic policy reuse in a reinforcement learning agent (2006, background): “Advice rules [7] define the actions to be preferred in different sets of states.”
Bridging the Gap: Converting Human Advice into Imagined Examples (2019, background): “RATLE (Maclin et al., 2005) was one of the first reinforcement learning systems to incorporate advice.”