search

Anmol6 / DNGO-BO

Bayesian optimization with DNGO (Deep Networks for Global Optimization)
11 stars 4 forks source link

test out the code on an actual problem #6

Open mgelbart opened 7 years ago

mgelbart commented 7 years ago

Pick a data set of your choice, and a model (like sklearn SVM or whatever) and actually try optimizing the hyperparameters. Make a plot of validation error vs. number of function evaluations. Also make a plot of running time vs. number of function evaluations, since the whole point of this is that it doesn't scale as badly as a GP.

mgelbart commented 7 years ago

Even before this, let's use a synthetic function with no noise in the function evaluations, and plot "best value found so far" vs. "number of iterations". Then do random search and put that as another curve on the plot. Just to sanity check that we're doing better.