RLIF: Interactive Imitation Learning as Reinforcement Learning @rail7462
RLIF: Interactive Imitation Learning as Reinforcement Learning  @rail7462
Uploaded December 2023 | Updated November 2026, 2 hours ago
We present a reinforcement learning algorithm that runs under DAgger-like assumptions, which can improve upon suboptimal experts without knowing ground-truth rewards.

RLIF: Interactive Imitation Learning as Reinforcement Learning
Jianlan Luo*, Perry Dong*, Yuexiang Zhai, Yi Ma, Sergey Levine

Project page: rlif-page.github.io
RLIF: Interactive Imitation Learning as Reinforcement LearningFastRLAP: A System for Learning High-Speed Driving via Deep RL and Autonomous Practicing (Summary)CS 182: Lecture 13: Part 2: NLPCS 285: Lecture 21, RL with Sequence Models & Language Models, Part 1CS 182: Lecture 21: Part 3: Meta-LearningCS 182: Lecture 4: Part 2: OptimizationCS 185/285 (Spring 2026): Lecture 16, Model-Based RL AlgorithmsCS 182: Lecture 20: Part 3: Adversarial ExamplesCS 182: Lecture 8: Part 3: Computer VisionCS 285: Lecture 21, Transfer Learning, Part 4CS 182: Lecture 21: Part 1: Meta-LearningCS 285: Andrea Zanette: Towards a Statistical Foundation for Reinforcement Learning
RAIL |

RLIF: Interactive Imitation Learning as Reinforcement Learning

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER