This repository has been archived by the owner on Jun 14, 2024. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 31
[WIP] The draft implementation of the new execution time distributions. #168
Draft
zhenwendai
wants to merge
16
commits into
develop
Choose a base branch
from
new_dist_impl
base: develop
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
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
marpulli
reviewed
Apr 24, 2019
marpulli
reviewed
Apr 24, 2019
marpulli
reviewed
Apr 24, 2019
The current type of distribution class in MXFusion is a "symbolic" notation for probabilistic model and variational posterior definition. When developing sophisticated inference algorithm, one often faces the need of creating some intermediate distribution objects that do not exist in model/posterior definition. In the implementation of current inference algorithms, those distributions are explicitly represented in terms of their parameters. This is not ideal for memory organization and code reusing. If there are distribution objects for execution, those intermediate distributions can be better organization in an object-oriented fashion. Many distribution computation functions can be implemented on top of those execution time distribution classes such as Kullback-Leiber divergence. A good example of such distribution classes is the distribution class in Tensorflow (tf.distrbution). | ||
|
||
## Proposed Changes | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Will you add a section here on how the proposed changes affect the existing Distribution classes? Also on how this will or could extend to encompass functions as well? We probably want to strive to keep the two Factors consistent when it comes to things like where runtime vs graphical code lives.
Add multivariate normal run time distribution
* restructure the folders by renaming mxfusion.components.dist_impl to mxfusion.runtime.distributions * The remaining adjustment * Extend multivariate normal with general broadcasting. * Update the naming of runtime distributions. * Remove unnecessary comments. * Temporally fix the test case failure. * Rename RunTime to Runtime.
Codecov Report
@@ Coverage Diff @@
## develop #168 +/- ##
===========================================
- Coverage 88.14% 87.28% -0.86%
===========================================
Files 80 87 +7
Lines 4057 4208 +151
Branches 691 704 +13
===========================================
+ Hits 3576 3673 +97
- Misses 307 358 +51
- Partials 174 177 +3
Continue to review full report at Codecov.
|
…distribution except GPs (#182) * restructure the folders by renaming mxfusion.components.dist_impl to mxfusion.runtime.distributions * The remaining adjustment * Extend multivariate normal with general broadcasting. * Update the naming of runtime distributions. * Remove unnecessary comments. * Temporally fix the test case failure. * Rename RunTime to Runtime. * bring in the changes from develop * Implement the runtime distribution for beta and gamma. * The changes for svgp regression. * Implement Bernoulli distribution * Implement Categorical. * Implement Sigmoid Bernoulli distribution. * Add back the testcase for Beta distribution. * Implement Laplace runtime distribution. * Implement the Dirichlet runtime distribution. * Implemented Wishart distribution. * Implement the pointmass runtime distribution. * Add the point mass runtime distribution file. * Updated the interface for normal distributions. * Modify the multivariate normal distribution. * Implement Multivariate normal with mean and precision. * Update test requirements. * Update test requirements. * bug fix for svgpregression module. * Fix the error in the docstring.
Merge develop into new_dist_impl
DGP implementation
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.