Gradient Free Optimizers: A collection of modern optimization methods in Python

Re:Gradient Free Optimizers: A collection of modern optimiza... by morelandjs on Sunday February 28, 2021 @10:46AM (#26294156)

Getting a lot of these errors when using the GPR. Seems to be an issue with the estimation of the noise level term. Scratching my head on this one. Anyone familiar with what's going on? I tried adding some random noise to the optimization function, and it had the same issue.

...python3.8/site-packages/sklearn/gaussian_process/kernels.py:402: ConvergenceWarning: The optimal value found for dimension 0 of parameter k2__noise_level is close to the specified lower bound 1e-05. Decreasing the bound and calling fit again may find a better value.

[ Parent | on HN ]

Re:Gradient Free Optimizers: A collection of modern optimiza... by simonblanke on Sunday February 28, 2021 @10:57AM (#26294223)

Yeah this is a warning from sklearns gaussian process regressor. Sklearn probably wants you to increase the "n_restarts_optimizer"-parameter of the gpr, but from my experience this warning does not correlate with bad results from Gradient-Free-Optimizers.

Sklearn warnings are often difficult to silence. Just google "sklearn suppress warnings" to get a code snippet for this problem.

[ Parent | on HN ]

Re:Gradient Free Optimizers: A collection of modern optimiza... by morelandjs on Sunday February 28, 2021 @11:12AM (#26294329)

thanks!

[ Parent | on HN ]