ConceptAI & ML
Imitation learning
Learning a policy or reward function from expert demonstrations.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Supervised learning5Machine learning2Behavioral cloning1Expert trajectories1Policy1Reinforcement learning basics1Robotics1
Opens up
Concepts that name this one as a prerequisite.
DAgger2Behavioral cloning1Behavioral cloning analysis1Causal confusion1Dagger (dataset aggregation)1Data augmentation in imitation learning1Distribution shift1Distributional shift1Generalization vs. extrapolation1Generative adversarial imitation learning1Human imitation learning1Low-exploitable policies1
Lecturers
6Each in their own terms, from the easiest level.
Emma Brunskill
A technique that reduces reinforcement learning to supervised learning by assuming access to input demonstrations of good policies.Russ Tedrake
Learning strategy where a robot is trained to replicate dexterous actions such as turning book pages.Sergey Levine
Learning by directly copying observed behavior or inferring intent from demonstrations.Leslie Kaelbling
A machine learning technique where an agent learns to perform a task by observing demonstrations from an expert.John Langford
Training a model by mimicking expert demonstrations, which can suffer from distribution shift when stepping off correct sequences.
Lectures that assume it
8- Imitation Learning II: Covariate Shift & Dataset Aggregation (DAgger)
- How Can Robots Learn to Interact and Coordinate with Others?
- Expanding What Robots Can Learn in the Real World
- Inverse Reinforcement Learning, Part 1
- Inverse Reinforcement Learning, Part 4
- Imitation Learning I: Behavioral Cloning & Distributional Shift
- Model-Based Reinforcement Learning with Policies
- Imitation Learning III: Inverse RL & Generative Adversarial Imitation Learning (GAIL)
Connections
23The lectures that define it or assume it, and the concepts next to it.
Concepts8
Show 2 moreShow fewer
Lectures15
- Introduction to Reinforcement Learning (CS234, 2024, Lecture 1)
- Robot Dynamics and Model-Based Control
- Sequential Decision-Making: Policy Gradients vs. Value-Based Architectures
- The Role of Rationality in Modern Robotics
- Guest Lecture: Dorsa Sadigh (UC Berkeley CS 285)
- Imitation Learning, Foundation Models, and Course Wrap-up
Show 9 moreShow fewer
- An Occam's Razor Principle for Transformers?
- Imitation Learning II: Covariate Shift & Dataset Aggregation (DAgger)
- How Can Robots Learn to Interact and Coordinate with Others?
- Expanding What Robots Can Learn in the Real World
- Inverse Reinforcement Learning, Part 1
- Inverse Reinforcement Learning, Part 4
- Imitation Learning I: Behavioral Cloning & Distributional Shift
- Model-Based Reinforcement Learning with Policies
- Imitation Learning III: Inverse RL & Generative Adversarial Imitation Learning (GAIL)