[KongchangAI]
ConceptExpected Improvement / EI

Expected Improvement

A classic acquisition function in Bayesian optimization. It computes the expected improvement a candidate point could achieve relative to the current best value after sampling, naturally balancing exploration and exploitation, and usually has a closed-form expression.

Timeline (last 90 days)

Oct 14

Standard Bayesian optimization acquisition functions (such as GP-UCB, EI, Thompson sampling) are almost all designed to minimize cumulative regret, and therefore naturally favor exploitation over exploration.

Unverified50%

All Facts (1)

Source Articles