Made a version of MlBayesim that doesn't store NaNs in the tables so that huge models can be estimated. #1750
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
Added a new implementation of the BayesIm interface, MlBayesImOld.java, for representing directed acyclic graphs in Bayes nets. This class also supports operations for manipulating the node tables including setting and retrieving probabilities, normalizing node tables, and checking if table rows are incomplete.
Implemented a new class ProbMap.java in bayes package, as a map between a unique integer index for a particular node and its probability. This mapping omits NaN values and it's implemented in MlBayesIm class. MlBayesIm uses either traditional probability matrices or these new probability maps, based on a flag setting.
Introduced ProbMap class, representing a efficient mapping between a unique index and node probability, neglecting NaN values. Integrated this into MlBayesIm replacing old matrix-based approach, governed by useProbMatrices flag.
Updated ProbMap.java and MlBayesIm.java with detailed Javadoc comments for better understanding of the code. Also, refined the copyDataToProbMatrices() method in MlBayesIm.java to consider the 'useProbMatrices' flag before copying data from 'probs' array to 'probMatrices' array.
The ProbMap class and its instances have been renamed to CptMap to better reflect their purpose of handling Conditional Probability Tables (CPTs). Changes were also executed in MlBayesIm class to reflect this modification. Added more detailed Javadoc comments for improved readability and understanding.
The ProbMap class and its instances have been renamed to CptMap to better reflect their purpose of handling Conditional Probability Tables (CPTs). Changes were also executed in MlBayesIm class to reflect this modification. Added more detailed Javadoc comments for improved readability and understanding.
The ProbMap class and its instances have been renamed to CptMap to better reflect their purpose of handling Conditional Probability Tables (CPTs). Changes were also executed in MlBayesIm class to reflect this modification. Added more detailed Javadoc comments for improved readability and understanding.
Fixed grammar in the explanatory comments of MlBayesIm.java, mainly focusing on the description of the division of labour among different classes and the purposes of the different dimensions in the data arrays. Also, improved sentence flow in a few places to enhance readability.
Updated MlBayesIm storage strategy from a three-dimensional array to an array of CptMap objects, which offer a more efficient representation for large conditional probability tables. Enhanced code readability by removing unnecessary checks and revising comments for clarity. This new method omits the storage of NaNs allowing space savings for sparse tables.
Transitioned to an efficient storage method for large conditional probability tables in the CptMap class. This change
Removed duplicative comments and unnecessary explanations about unique integer index calculation in CptMap class. Refocused on the key functionality: storage of large conditional probability tables in a compact form excluding NaN values.
Modified the explanation of MlBayesIm's probability storage method from a three-dimensional array to a sparse method that does not store NaNs. This change highlights the method's efficiency when working with large Bayesian probabilistic models. The old storage method's description remains for backward compatibility.
Removed an embedded class called SimulationTask along with related code lines in MlBayesIm.java file. The modification simplifies the constructSample function, as the simulation now utilizes a sequential approach rather than a fork/join parallel computation, making it more straightforward.
Refactored the MlBayesIm.java file to replace the use of probability matrices with CptMaps for storing probabilities. This change introduces more efficient data structures and methods for maintaining and accessing probabilities. Backward compatibility is maintained through the fallback to a "probs" array if CptMaps are not used.
Modified MlBayesIm.java to implement CptMaps instead of probability matrices for better memory management. While this update optimizes probability storage and access methods, it also ensures backward compatibility by keeping the old method and introducing a flag to select the method.
This commit simplifies the getVariableNames method in the MlBayesIm class. It replaces the for loop that used indices with a more straightforward enhanced for loop that directly iterates over the nodes. Consequently, this makes the code cleaner and easier to understand without affecting functionality.
Null check has been added in the existsParameterizedConstructor method in SessionNode class. If the model class is null, then the method will return false indicating that no constructor exists. This added validation is intended to prevent potential issues related to null object handling.
Refactored the conditional check in `RandomGraph.java` to use `isEmpty()` function instead of `size() == 0`. Adjusted parameter names in `BayesPmWrapper.java` for improved clarity. Corrected a typographical error in `RandomGraphEditor.java` from "graphRandomFoward" to "graphRandomForward". These changes improve code readability and maintainability.
Corrected a typographical error in a comment within the `Statistic.java` file. This edit improves the readability of the codebase for future developers by ensuring comments accurately describe the intended functionality and behavior of the code.
Updated the usage of initialization methods in various Bayes related classes, specifically changing from "MlBayesIm.RANDOM" and "MlBayesIm.MANUAL" to "MlBayesIm.InitializationMethod.RANDOM" and "MlBayesIm.InitializationMethod.MANUAL". Created new files for CptMapProbs and CptMapCounts as part of the refactoring. This update helps make the code more readable and consistent across different parts of the application.
Updated the usage of initialization methods in various Bayes related classes, specifically changing from "MlBayesIm.RANDOM" and "MlBayesIm.MANUAL" to "MlBayesIm.InitializationMethod.RANDOM" and "MlBayesIm.InitializationMethod.MANUAL". Created new files for CptMapProbs and CptMapCounts as part of the refactoring. This update helps make the code more readable and consistent across different parts of the application.
Updated the usage of initialization methods in various Bayes related classes, specifically changing from "MlBayesIm.RANDOM" and "MlBayesIm.MANUAL" to "MlBayesIm.InitializationMethod.RANDOM" and "MlBayesIm.InitializationMethod.MANUAL". Created new files for CptMapProbs and CptMapCounts as part of the refactoring. This update helps make the code more readable and consistent across different parts of the application.
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.
This was a wish list item.