Reinforcement Learning with Human Teachers: Evidence of Feedback and Guidance with Implications for Learning Performance

By Andrea Lockerd Thomaz, Cynthia Lynn Breazeal. 2006

313 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

Widening the Pipeline in Human-Guided Reinforcement Learning with Explanation and Context-Aware Data Augmentation (2020, background): “As an early attempt, Thomaz et al. allow humans to give anticipatory guidance rewards and point out the object tied to the reward [44].”
Interactive Learning from Policy-Dependent Human Feedback (2017, background): “Indeed, much human-centered RL work has investigated and employed different models of human-feedback (Knox & Stone, 2009b; Thomaz & Breazeal, 2006; 2007; 2008; Grifﬁth et al., 2013; Loftin et al., 2015).”
Deep TAMER: Interactive Agent Shaping in High-Dimensional State Spaces (2017, background): “Most closely related to the work presented here, several methods have recently been proposed that allow an agent to learn from interaction with a non-expert human (Thomaz and Breazeal 2006; MacGlashan et al. 2014; Loftin et al. 2014; 2016; Peng et al. 2016).”
Safe Reinforcement Learning via Shielding (2017, background): “In each time step, the human teacher tunes the reward signal before sending it to the agent (Thomaz and Breazeal 2006; 2008).”
Learning Hierarchical Teaching Policies for Cooperative Agents (2019, background): “learning agent to follow. Curriculum learning has many applications, including recent works that learn a training data curriculum for an image classifier [10, 16]. Another work of Thomaz and Breazeal [35] studies transferring human knowledge to an RL agent via providing the reward signals. While mos”
A survey of robot learning from demonstration (2009, unknown): “A modified RL algorithm learns task reward entirely based on individual rewards interactively provided by a human user [80].”
Deep Reinforcement Learning from Policy-Dependent Human Feedback (2019, unknown): “While there have been varied approaches to tackling HRL problems (Thomaz & Breazeal, 2006; Knox & Stone, 2008; Grifﬁth et al., 2013; Loftin et al., 2015; Knox & Stone, 2010; Vien & Ertel, 2012), we extend an approach that is exceptionally considerate of how human trainers provide feedback to…”
FRESH: Interactive Reward Shaping in High-Dimensional State Spaces using Human Feedback (2020, background): “Reward shaping modifies rewards supplied by the environment to accelerate the learning process of the agent [5, 17, 41].”
Curriculum Learning for Reinforcement Learning Domains: A Framework and Survey (2020, methodology): “Published research on Interactive Reinforcement Learning (Thomaz and Breazeal, 2006; Knox and Stone, 2009; Suay and Chernova, 2011; Knox and Stone, 2012; Griffith et al., 2013; Subramanian et al., 2016; Loftin et al., 2016; MacGlashan et al., 2017) has shown that RL agents can successfully speed up ”
Expressing Arbitrary Reward Functions as Potential-Based Advice (2015, background): “On the other hand, there has been a lot of research in human-provided advice (Thomaz and Breazeal 2006), (Knox et al. 2012).”
TAMER: Training an Agent Manually via Evaluative Reinforcement (2008, background): “Thomaz and Breazeal [6] interfaced a human trainer with a table-based, Q-learning agent in a relatively simple environment .”
How Humans Teach Agents - A New Experimental Perspective (2012, background): “This finding confirms observations by Thomaz and Breazeal [37].”
Reinforcement Learning from Demonstration through Shaping (2015, background): “Other authors do not consider demonstrations, but allow a human to interactively provide reward or advice during agent learning [Thomaz and Breazeal, 2006; Knox and Stone, 2010].”
Challenges in Shared-Environment Human-Robot Collaboration (2013, background): “One technique that particularly takes advantage of humans in the loop is the combination of socially guided machine learning and active learning [8], [9].”
Interactively shaping robot behaviour with unlabeled human instructions (2019, background): “Moreover, previous work has shown that, in addition to evaluative feedback, human teachers want to provide guidance about future actions [44].”