Bayesian Optimization: Lower Confidence Bound
Beginner
The problem, from first principles
A simple acquisition rule is to try the point with the lowest optimistic estimate: the predicted mean minus kappa times its uncertainty. Points with high uncertainty look better, which drives exploration.
From theory to code
Implement lower_confidence_bound(mu, sigma, kappa), returning mu - kappa * sigma.
Constraints
- For minimization, pick the candidate with the smallest value.
Hints
Hint 1
Subtract kappa times sigma from mu, element-wise.
Keywords
research-papersclassical-mlbayesian-optimizationhyperparameters
Python
Run your code to see results
No results yet
Run your code to check it against the examples, or submit to run every test.