Implement TFP's base measure interface for distrax dists and bijectors. #104
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
![copybara-service[bot]](https://avatars.githubusercontent.com/in/44061?s=60&v=4){kind=small}
copybara-service
bot
force-pushed
the
test_427808598
branch
3 times, most recently
from
069a056
to
d864f08
Compare
copybara-service
bot
force-pushed
the
test_427808598
branch
2 times, most recently
from
6c0f52f
to
52dfd83
Compare
copybara-service
bot
force-pushed
the
test_427808598
branch
2 times, most recently
from
d48c4de
to
3c15ce1
Compare
The interface needs two new methods: an `experimental_local_measure` on `Distribution`s, and an `experimental_density_correction` on `Bijector`s. We implement these methods using the default assumptions already made by TFP: that the tangent space to a manifold has the standard basis on R^n, and that the transformations on this manifold are dimension-preserving. In the future, TFP will define more appropriate bases and density corrections for subclasses of `Distribution` and `Bijector`; distrax is not expected to follow this protocol. See: - `tfp.python.experimental.tangent_spaces` - `tfp.distributions.experimental_local_measure` - `tfp.bijectors.experimental_compute_density_correction` - Radul and Alexeev, AISTATS 2021, “The Base Measure Problem and its Solution”, https://arxiv.org/abs/2010.09647. PiperOrigin-RevId: 431515351
copybara-service
bot
force-pushed
the
test_427808598
branch
from
3c15ce1
to
5fa52d1
Compare
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Outdated suggestions cannot be applied.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.
Implement TFP's base measure interface for distrax dists and bijectors.
The interface needs two new methods: an
experimental_local_measureonDistributions,and an
experimental_density_correctiononBijectors. We implement these methodsusing the default assumptions already made by TFP: that the tangent space to a manifold
has the standard basis on R^n, and that the transformations on this manifold are
dimension-preserving. In the future, TFP will define more appropriate bases and
density corrections for subclasses of
DistributionandBijector; distrax is not expectedto follow this protocol.
See:
tfp.python.experimental.tangent_spacestfp.distributions.experimental_local_measuretfp.bijectors.experimental_compute_density_correction