Investigating Simple Object Representations in Model-Free Deep Reinforcement Learning
By Guy Davidson, Brenden M. Lake. 2020
11 citations recorded by the source corpus.
We explore the benefits of augmenting state-of-the-art model-free deep reinforcement algorithms with simple object representations. Following the Frostbite challenge posited by Lake et al. (2017), we identify object representations as a critical cognitive capacity lacking from current reinforcement learning agents. We discover that providing the Rainbow model (Hessel et al.,2018) with simple, feature-engineered object representations substantially boosts its performance on the Frostbite game from Atari 2600. We then analyze the relative contributions of the representations of different types of objects, identify environment states where these representations are most impactful, and examine how these representations aid in generalizing to novel situations.
Subfields: Artificial Intelligence
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
- Sjoerd van Steenkiste
- J. Urgen Schmidhuber
- Klaus Greff
- Yoshua Bengio
- Sergey Levine
- Quoc V. Le
- Mohit Bansal
- Kyunghyun Cho
Citations in the corpus
- On the Binding Problem in Artificial Neural Networks (2020, background): “Countless other variations that vary in terms of the division of ur between the symbolic and neural components and the choice of a mechanism used to couple them are possible (McGarry et al., 1999; Davidson and Lake, 2020).”