Learning non-myopically from human-generated reward

37 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.

Deep Reinforcement Learning from Human Preferences (2017, background): “TAMER (Knox, 2012; Knox and Stone, 2013) also learn a reward function, however they consider much simpler settings where the desired policy can be learned relatively quickly.”
Interactive Learning from Policy-Dependent Human Feedback (2017, methodology): “We chose TAMER as a comparison because, to our knowledge, it is the only HCRL algorithm with success on a similar platform (Knox et al., 2013).”
Power to the People: The Role of Humans in Interactive Machine Learning (2014, background): “…however, has started to explore new interfaces, including interfaces for letting users label parts of items (e.g., Fails & Olsen 2003), adjust model parameters and cost functions (e.g., Kapoor et al. 2010), and directly modify a classifier’s features (e.g., Kulesza et al. 2011).”
DQN-TAMER: Human-in-the-Loop Reinforcement Learning with Intractable Feedback (2018, methodology): “, we employ so-called reward shaping [28], [29] to incorporate human feedback.”
Improved human–robot team performance through cross-training, an approach inspired by human team training practices (2015, background): “Knox and Stone wrote that the general positivity of human reward can actually lead to “positive circuits” (Knox and Stone, 2013): repeatable sequences of behavior that lead to accumulated reward values higher than that of the goal state.”