Bayesian Optimization in Action MEAP V06 by meap-version-6.html

Author:meap-version-6.html
Language: eng
Format: epub, pdf

A point that is far away, which we don’t know much about, might give a lower improvement probability, but because there’s a chance that this point will give a large improvement, EI might assign a higher score to it. In other words, while PoI might be considered a risk-averse BayesOpt policy that cares about improving from the incumbent, however small the improvement is, EI balances between the risk and reward in order to find the point that best balances the tradeoff. This is illustrated in Figure 4.16, which compares PoI and EI using the same data set and trained GP.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.