Re:Show HN: A radically more scalable alternative to Bayesia...
by Straw on Saturday January 27, 2018 @05:41AM
(#16245691)
Gaussian processes are only one implementation of Bayesian optimization- afaict, you have created another using a different prior which has nicer computational properties, mostly due to assuming the function has little global structure, a reasonable one for tricky optimization problems. I wonder how it would compare to a GP approximation using a space-partitioning data structure, only considering nearby points when calculating sample utility.
[ Parent
| on HN ]
Re:Show HN: A radically more scalable alternative to Bayesia...
by chroem- on Saturday January 27, 2018 @09:21PM
(#16249374)
Hi there,
While it shares a lot of similarities with conventional Bayesian Optimization, I was reluctant to call it BO since it doesn't have a concept of explicit error bounds.