Active reward learning with a novel acquisition function

By Christian Daniel, Oliver Kroemer, Malte Viering, Jan Metz, Jan Peters. 2015

69 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Control and Systems Engineering, Reinforcement 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

1. Sergey Levine
2. Pieter Abbeel
3. Chelsea Finn
4. Jan Peters
5. Gerhard Neumann
6. Yoshua Bengio
7. Trevor Darrell
8. Jan Leike

In Control and Systems Engineering

1. Sergey Levine
2. Pieter Abbeel
3. Jan Peters
4. Gerhard Neumann
5. J. Bagnell
6. Jens Kober
7. Chelsea Finn
8. Jan Leike

In Reinforcement Learning

1. Sergey Levine
2. Pieter Abbeel
3. Chelsea Finn
4. Jan Peters
5. Gerhard Neumann
6. J. Bagnell
7. Jan Leike
8. Shane Legg

Citations in the corpus

• Deep Reinforcement Learning from Human Preferences (2017, background): “…reinforcement learning from human ratings or rankings, including Akrour et al. (2011), Pilarski et al. (2011), Akrour et al. (2012), Wilson et al. (2012), Sugiyama et al. (2012), Wirth and Fürnkranz (2013), Daniel et al. (2015), El Asri et al. (2016), Wang et al. (2016), and Wirth et al. (2016).”
• Reward learning from human preferences and demonstrations in Atari (2018, background): “…learning from human ratings or rankings (Wirth et al., 2017): Knox and Stone (2009), Pilarski et al. (2011), Akrour et al. (2012), Wilson et al. (2012), Wirth and Fürnkranz (2013), Daniel et al. (2015), El Asri et al. (2016), Wirth et al. (2016), Mathewson and Pilarski (2017), and others.”
• An Algorithmic Perspective on Imitation Learning (2018, methodology): “With active inverse reinforcement learning we refer to learning the reward function when the robot is able to influence the demonstrations [Daniel et al., 2015].”
• Learning Interactively to Resolve Ambiguity in Reference Frame Selection (2020, background): “Similar analysis on user-friendly interactions can be found in the literature about active reward learning [14], reward learning through preferences [15] and designing robots that asks good questions [16].”
• Sample and Feedback Efficient Hierarchical Reinforcement Learning from Human Preferences (2018, methodology): “Perhaps the closest work to ours is that of [21], which also learns to grasp from human feedback.”