mlsanityData to 14 Oct 2026

SKILL-IL: Disentangling Skill and Knowledge in Multitask Imitation Learning

arXiv 2205.03130 ↗First posted 6 May 2022cs.LGcited by 4

On the map: Offline RL and imitation learning from data → Robot manipulation imitation learning

Abstract (arXiv)

In this work, we introduce a new perspective for learning transferable content in multi-task imitation learning. Humans are able to transfer skills and knowledge. If we can cycle to work and drive to the store, we can also cycle to the store and drive to work. We take inspiration from this and hypothesize the latent memory of a policy network can be disentangled into two partitions. These contain either the knowledge of the environmental context for the task or the generalizable skill needed to solve the task. This allows improved training efficiency and better generalization over previously unseen combinations of skills in the same environment, and the same task in unseen environments. We used the proposed approach to train a disentangled agent for two different multi-task IL environments. In both cases we out-performed the SOTA by 30% in task success rate. We also demonstrated this for navigation on a real robot.

Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Placed in Robot manipulation imitation learning, one of 13 directions in Offline RL and imitation learning from data. Builds on 15 of its 26 references that are in the map. First posted on arXiv 6 May 2022.

Builds on

15 of 26 references.

References · 15

  1. 2013Distributed Representations of Words and Phrases and their Compositionalitycited by 3,474
  2. 2015FaceNet: A Unified Embedding for Face Recognition and Clusteringcited by 2,849
  3. 2018Learning deep representations by mutual information estimation and maximizationcited by 1,273
  4. 2017Variational Inference of Disentangled Latent Concepts from Unlabeled Observationscited by 241
  5. 2017Distral: Robust Multitask Reinforcement Learningcited by 233
  6. 2016Modular Multitask Reinforcement Learning with Policy Sketchescited by 203
  7. 2019Diagnosing and Enhancing VAE Modelscited by 193
  8. 2018Isolating Sources of Disentanglement in Variational Autoencoderscited by 167
  9. 2017Zero-Shot Task Generalization with Multi-Task Deep Reinforcement Learningcited by 98
  10. 2017Hierarchical and Interpretable Skill Acquisition in Multi-task Reinforcement Learningcited by 77
  11. 2018Disentangling Latent Space for VAE by Label Relevant/Irrelevant Dimensionscited by 19
  12. 2020NestedVAE: Isolating Common Factors via Weak Supervisioncited by 13
3 more
  1. 2018Hyperprior Induced Unsupervised Disentanglement of Latent Representationscited by 6
  2. 2019Gated Variational AutoEncoders: Incorporating Weak Supervision to Encourage Disentanglementcited by 6
  3. 2021Robot in a China Shop: Using Reinforcement Learning for Location-Specific Navigation Behaviourcited by 2

Built on by

Cited by year

Citing papers per year: 2022: 1; 2023: 2; 2024: 1.

4 papers.

  1. 2023Multi-task Hierarchical Adversarial Inverse Reinforcement Learningcited by 15
  2. 2023PLEX: Making the Most of the Available Data for Robotic Manipulation Pretrainingcited by 9
  3. 2022Generalizing to New Tasks via One-Shot Compositional Subgoalscited by 1
  4. 2024Shared-unique Features and Task-aware Prioritized Sampling on Multi-task Reinforcement Learningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2022Learning and Retrieval from Prior Data for Skill-based Imitation Learning
  2. 2022Latent Plans for Task-Agnostic Offline Reinforcement Learning
  3. 2021Intrinsically Motivated Open-Ended Multi-Task Learning Using Transfer Learning to Discover Task Hierarchy
  4. 2026Dual-Process Atomic Skill Learning: Decoupling Semantic Reasoning and Real-Time Control
  5. 2020Scalable Multi-Task Imitation Learning with Autonomous Improvement
  6. 2024Rethinking Mutual Information for Language Conditioned Skill Discovery on Imitation Learning
  7. 2026Offline Discovery of Interpretable Skills from Multi-Task Trajectories
  8. 2023CasIL: Cognizing and Imitating Skills via a Dual Cognition-Action Architecture