- Using 10 materials for one Model?
- Unity 2D How to prevent character from double jump
- Synchronising movement across network in Unity RTS
- How can I get the force of a collision in Love2D?
- Screen record in unity3d
- Limit of hits in Google Analytics when using Android SDK
- Is there an SEO reason why I shouldn't redirect the root domain to a subdomain?
- How can i play Animal Crossing: New Leaf Welcome Amiibo Multiplayer?
- 4 Ender Dragons on Outer Islands?
- In path of exile, I sometimes die well after a fight is over
- How can one remove lag spikes due to memory dumps?
- How can I tweak to Heatran in Pokemon Diamond?
- Why did Blackbox tell me “The game isn’t designed to be played like that”
- Building circuits with BuildCraft
- How do you move chests in your house?
- How can I undo “hide all from
”? (NOT hide all from person, page, group, or app)
- Trying to set interests on campaign details to boost a post on Facebook
- Sent items not saved in sent items folder in Outlook 365
- Does Gmail out-of-office/vacation responder reply to other addresses in Cc?
- Badge notification icon color
Deep Learning: What are the differences between DeepMind's Learning to Learn method and a grid search of a network's hyperparameters?
If we have a meta-learner that trains an optimizer (who contains certain hyperparameters) and the optimizer is fine-tuned by the meta-learner depending on how it performs, how is it different from a usual grid search of the best hyperparameters? One way I think it's different is that the meta-learner can supposedly 'intelligently' find the best optimizer but a gridsearch is rather a brute-force method. But a grid search would likely include human knowledge on the range of hyperparameters where the model is likely to perform well.
My current impression is that the meta-learner simply tweaks the hyperparameters of the optimizer (which are usually fixed in many other cases) after certain number of epochs, evaluates the performance, then dynamically changes how the hyperparameters should be tweaked. Is this what the authors of the paper did?