ConceptAI & ML
Distributional shift
The problem where the distribution of states encountered during test time differs from the distribution of states seen in the training data.
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.
Reinforcement learning2State distributions2Supervised learning2Behavioral cloning1Expected value1Imitation learning1Machine learning models1Neural networks1
Opens up
Concepts that name this one as a prerequisite.
Lecturers
2Each in their own terms, from the easiest level.
Sergey Levine
Sara Beery
The mismatch between anticipated conditions or training data distributions and the actual deployment environment.
Lectures that assume it
1Connections
14The lectures that define it or assume it, and the concepts next to it.
Concepts6
Lectures8
- Imitation Learning II: Covariate Shift & Dataset Aggregation (DAgger)
- Fundamental concepts in deep reinforcement learning
- Offline Reinforcement Learning II
- Generalization: Out-of-Distribution (OOD)
- Exploration of advanced topics in deep reinforcement learning
- Model-Based Reinforcement Learning with Policies