Problem submission - Learning gradient descent with synthetic objectives #10
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
farizrahman4u
suggested changes
Nov 12, 2016
| ## Problem description | ||
| Current optimization algorithms for neural networks such as SGD, RMSProp and Adam are hand-crafted and generally quite simple. This can be partly explained by the high-dimensional, non-convex nature of neural network's objective functions which human intuition, normally limited to three spatial dimensions, is not well-suited for. A learning algorithm, therefore, may be able to design a superior optimizer. | ||
|
|
||
|
|
| ## Project status | ||
| A formula for generating synthetic objective functions has been created. These functions are differentiable and their dimensionality and degree of non-linearity can be controlled with hyperparameters. | ||
|
|
||
|
|
farizrahman4u
approved these changes
Nov 13, 2016
|
I would like to see more people reviewing this proposal and giving feedback. |
|
@cjratcliff Do you have links for the stuff mentioned under project status? For the optimizers trained with supervision and reinforcement learning? |
|
@farizrahman4u Sure. I've now uploaded the code to Github here. Note that the learned optimizers can't be used as seamlessly as TensorFlow's inbuilt ones right now. Gradients have to be explicitly calculated and passed into the optimizer, which outputs the updates. Don't expect anything much from the RL version. It's far enough from being complete that it's no better than choosing random updates. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.