You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello @bowenbaker , @peterldowns
firstly, thank your for the interesting paper and repository! :)
This is maybe rather a question for the paper, not sure if it belong here - feel free to move it.
Do you have any plans to release a comparison of your method with recent methods for Bayesian Optimization / genetic optimization and the like, as part of current or future work? For example using methods from black box optimization competition or this one for bayesian optimization: Snoek 2015. In latter, it seems that authors get a bit better results with CIFAR10 and similar results with CIFAR100, but I am not sure whether this is due to different optimization algorithm used or due to specific ranges of hyperparameters / architectures etc.
I think it would be pretty interesting to see such a comparison. What do you think?
The text was updated successfully, but these errors were encountered:
Hello @bowenbaker , @peterldowns
firstly, thank your for the interesting paper and repository! :)
This is maybe rather a question for the paper, not sure if it belong here - feel free to move it.
Do you have any plans to release a comparison of your method with recent methods for Bayesian Optimization / genetic optimization and the like, as part of current or future work? For example using methods from black box optimization competition or this one for bayesian optimization: Snoek 2015. In latter, it seems that authors get a bit better results with CIFAR10 and similar results with CIFAR100, but I am not sure whether this is due to different optimization algorithm used or due to specific ranges of hyperparameters / architectures etc.
I think it would be pretty interesting to see such a comparison. What do you think?
The text was updated successfully, but these errors were encountered: