AI engineering glossary · Reinforcement learning
What is exploitation?
Exploitation: Choosing the option that currently looks best based on what has been learned so far, to collect as much score as possible now. Too much of it stops learning, since alternatives are never tested.
Exploitation: an example
Always recommending the song with the highest past play count, without trying any new tracks.
Related terms
More Reinforcement learning terms
- Exploration
- Exploration-exploitation trade-off
- Function approximation
- Goal-conditioned reinforcement learning
- Greedy policy
- Hierarchical reinforcement learning
- Hindsight experience replay
- Horizon
Where you learn it
Exploitation is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.