search

rmcantin / bayesopt

BayesOpt: A toolbox for bayesian optimization, experimental design and stochastic bandits.
GNU Affero General Public License v3.0
396 stars 92 forks source link

noise effect #21

Closed zhaozhongch closed 6 years ago

zhaozhongch commented 6 years ago

About noise the instruction says: Too much noise results in slow convergence while not enough noise might result in not converging at all Why is that?

rmcantin commented 6 years ago

We use a GP for modeling the function to optimize. If you asume 0 noise, you are considering the GP is a perfect model, but of course, the GP cannot model any function. If you asume a large noise, you are considering that the GP is a bad model, thus it does not help to converge.