Add Feedback Alignment implementation #848
Open
Conversation
This file contains hidden or 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
Implements biologically plausible learning using fixed random feedback matrices instead of weight transposes. Addresses the weight transport problem mentioned in issue google-deepmind#750. Features: - FALinear module compatible with standard Haiku layers - Custom VJP for feedback-based gradient computation - Comprehensive test suite - Google-style documentation Reference: Lillicrap et al. (2016) Nature Communications
|
Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). View this failed invocation of the CLA check for more information. For the most up to date status, view the checks section at the bottom of the pull request. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
Summary
This PR implements Feedback Alignment (FA) for dm-haiku, addressing the biological plausibility concerns raised in issue #750.
Background
Feedback Alignment is a biologically plausible alternative to standard backpropagation that solves the weight transport problem by using fixed random feedback matrices instead of weight transposes during the backward pass.
Implementation
Why FA instead of DFA?
The original goal was to implement Direct Feedback Alignment (DFA). However, after extensive investigation, we discovered that DFA is incompatible with JAX's autodiff architecture due to shape mismatch constraints in custom_vjp. FA provides the same biological plausibility benefits while working seamlessly with JAX/Haiku.
Key Features
Testing
All 5 test cases pass successfully:
References
Lillicrap, T. P., et al. (2016). Random synaptic feedback weights support error backpropagation for deep learning. Nature Communications, 7, 13276.
Closes #750