Intrinsically Motivated Reinforcement Learning: A Promising Framework for Developmental Robot Learning

By Andrew Stout, George Dimitri Konidaris, Andrew Barto. 2005

79 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Mechanical Engineering, Reinforcement Learning, Robotics & Control

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. George Dimitri Konidaris
2. Pierre-Yves Oudeyer
3. Gianluca Baldassarre
4. Marco Mirolli
5. Andrew Barto
6. Bruno Castro da Silva
7. Massimiliano Schembri
8. Romain Laroche

In Mechanical Engineering

1. Gianluca Baldassarre
2. Andrew Barto
3. George Dimitri Konidaris

In Reinforcement Learning

1. George Dimitri Konidaris
2. Marco Mirolli
3. Bruno Castro da Silva
4. Gianluca Baldassarre
5. Pierre-Yves Oudeyer
6. Andrew Barto
7. Harm van Seijen
8. Sergey Levine

In Robotics & Control

1. Andrew Barto
2. George Dimitri Konidaris
3. Gianluca Baldassarre
4. Bruno Castro da Silva
5. Marco Mirolli
6. Jan Peters

Citations in the corpus

• Learning Parameterized Skills (2012, background): “, by Harlow [24]) and in the machine learning and developmental robotics communities [60, 61, 67, 65].”
• Evolution and Learning in an Intrinsically Motivated Reinforcement Learning Robot (2007, background): “A key point is that the system uses the prediction error of the option model as an internal reward to decide which option to invoke: the effect is that until the ability to produce the associated ‘salient event’ is not refined, an option continues to generate internal rewards and hence to be selecte”
• Hybrid Reward Architecture for Reinforcement Learning (2017, background): “Intrinsic motivation (Stout et al., 2005; Schmidhuber, 2010) uses the above observation to improve learning in sparse-reward domains.”
• Evolving Childhood's Length and Learning Parameters in an Intrinsically Motivated Reinforcement Learning Robot (2007, background): “…the most promising frameworks that has been proposed to this purpose is 'Intrinsically Motivated Reinforcement Learning (IMRL)' (Barto et al., 2003; Stout et al., 2005) IMRL is based on the idea that natural organisms, especially the most sophisticated ones like humans and primates, are not driven…”