Imitation Learning
Imitation learning aims to train agents to mimic expert behavior by learning from observational data, primarily focusing on efficiently transferring complex skills from humans or other advanced controllers to robots. Current research emphasizes improving data efficiency through techniques like active learning, data augmentation, and leveraging large language models to provide richer context and handle failures. This field is crucial for advancing robotics, autonomous driving, and other areas requiring complex control policies, as it offers a more data-driven and potentially less labor-intensive approach than traditional programming methods.
Papers
Confidence-Based Skill Reproduction Through Perturbation Analysis
Brendan Hertel, S. Reza Ahmadzadeh
CCIL: Context-conditioned imitation learning for urban driving
Ke Guo, Wei Jing, Junbo Chen, Jia Pan
An Imitation Learning Based Algorithm Enabling Priori Knowledge Transfer in Modern Electricity Markets for Bayesian Nash Equilibrium Estimation
Ziqing Zhu, Ka Wing Chan, Siqi Bu, Ze Hu, Shiwei Xia
Get Back Here: Robust Imitation by Return-to-Distribution Planning
Geoffrey Cideron, Baruch Tabanpour, Sebastian Curi, Sertan Girgin, Leonard Hussenot, Gabriel Dulac-Arnold, Matthieu Geist, Olivier Pietquin, Robert Dadashi
CALM: Conditional Adversarial Latent Models for Directable Virtual Characters
Chen Tessler, Yoni Kasten, Yunrong Guo, Shie Mannor, Gal Chechik, Xue Bin Peng
Using Offline Data to Speed-up Reinforcement Learning in Procedurally Generated Environments
Alain Andres, Lukas Schäfer, Esther Villar-Rodriguez, Stefano V. Albrecht, Javier Del Ser
Behavior Retrieval: Few-Shot Imitation Learning by Querying Unlabeled Datasets
Maximilian Du, Suraj Nair, Dorsa Sadigh, Chelsea Finn