AI engineering glossary · Reinforcement learning
What is natural policy gradient?
Natural policy gradient: A policy update that follows the gradient rescaled by the inverse Fisher information matrix, so that steps are measured in how much the action distribution changes rather than in raw parameter distance. It is invariant to how the rule is parameterised and leads to trust-region methods.
Natural policy gradient: an example
A small change in parameters may barely change behaviour in one direction and a lot in another; this rescaling equalises them.
Related terms
More Reinforcement learning terms
- Noisy networks
- Off-policy
- Off-policy evaluation
- Offline reinforcement learning
- On-policy
- Optimal policy
- Options framework
- Overestimation bias
Where you learn it
Natural policy gradient is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.