CompILE: Compositional Imitation Learning and Execution

By Thomas Kipf, Yujia Li, Hanjun Dai, Vinícius Flores Zambaldi, Alvaro Sanchez-Gonzalez, Edward Grefenstette, Pushmeet Kohli, Peter W. Battaglia. 2018

139 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

Learning to Generalize Across Long-Horizon Tasks from Human Demonstrations (2020, background): “[21], which recovers subtasks from demonstrations in the spirit of hierarchical RL [39, 1].”
Relay Policy Learning: Solving Long-Horizon Tasks via Imitation and Reinforcement Learning (2019, methodology): “These methods aim to perform imitation learning by learning low-level primitives [9, 26] or latent conditioned policies [23] which meaningfully segment the demonstrations.”
Discovering Motor Programs by Recomposing Demonstrations (2020, methodology): “…spirit to these works, we too seek to learn abstractions from a given set of demonstrations, however unlike DDO (Fox et al., 2017), DDCO (Krishnan et al., 2017), and CompILE (Kipf et al., 2019), we can learn primitives beyond a discrete set of options in a relatively high dimensional domain.”
Modeling Long-horizon Tasks as Sequential Interaction Landscapes (2020, background): “To address this issue, there has a been a number of hierarchical imitation learning approaches [16, 53, 14, 29] that focus on segmenting long-horizon tasks into subcomponents.”
Learning Robot Skills with Temporal Variational Inference (2020, methodology): “The CompILE framework (Kipf et al., 2019) makes use of continuous latent variables to parameterize options, but requires carefully designed attention mechanisms, and is evaluated in relatively low-dimensional domains.”
Contrastive Learning of Structured World Models (2019, background): “…bind individual objects to slots or object ﬁles (Kahneman & Treisman, 1984; Kahneman et al., 1992), such as in the style of dynamic routing (Sabour et al., 2017), iterative inference (Greff et al., 2019; Engelcke et al., 2019) or sequential masking (Burgess et al., 2019; Kipf et al., 2019).”
Continuous Control with Action Quantization from Demonstrations (2021, background): “, 2020), or learned from interactions with the environment (Manschitz et al., 2015; Kipf et al., 2019; Shankar et al., 2019), or hand specified (Pastor et al.”
Parrot: Data-Driven Behavioral Priors for Reinforcement Learning (2020, background): “…et al., 2019), and methods that learn temporally extended actions, also known as options , from demonstrations, and then recompose them to perform long-horizon tasks through RL or planning (Fox et al., 2017; Krishnan et al., 2017; Kipf et al., 2019; Shankar et al., 2020; Shankar & Gupta, 2020).”