exploration-exploitation trade-off

Terms from Artificial Intelligence: humans at the heart of algorithms

Page numbers are for draft copy at present; they will be replaced with correct numbers when final book is formatted. Chapter numbers are correct and will not change now.

When interacting with the world in reinforcement learning, an agent has to choose whether to take the best action based on its existing knowledge (exploitation) or to try new thinsg in order to expand its knowledge (exploration). The former is a low risk option, but may miss longer-term gains -- an exmaple of a local minimum}. The agent therefore needs {meta-heuristics}} in order to manage this trade-off.

Used on Chap. 15: page 362; Chap. 16: pages 378, 379, 388; Chap. 22: page 547

Also known as exploitation, exploration