JBoost Code

99 lines (73 with data), 2.6 kB

JBoost Release Notes
--------------------

+ Release 2.2 + 
=================
 - Fix possible bugs with ADD_ROOT
 - Remove support for MultiLabel and MultiClass from java
 - Add more test cases
 - Add BrownBoost

+ Release 2.0 + 
=================
 - RobustBoost is implemented
 - Improved the score visualizer
 - Deprecated BrownBoost, YadaBoost
 - Replaced multiclass wrapper in JBoost with a python script


+ Release 1.4 + 
=================

 - NormalBoost is implemented
    - BrownBoost-like update
    - Asymetric cost
    - Iteration order evalution of the ADTree
 - More bugs fixed
 - Scripts updated and made more general

+ Release 1.3.1 + 
=================

 - BrownBoost now able to handle hypotheses that abstain
 - Several bug fixes (BrownBoost, Controller, etc)


+ Release 1.3 + 
===============

 - BrownBoost released 
 - Added a few README files
 - Fixed path bug in jboost wrapper and nfold.py
 - Minor improvements to margin.py and error.py
 - Added options outputting serialized trees
 - Added some more misc documentation


+ Release 1.2 + 
===============

 - Added a few README files
 - Added a lot of documentation
      - There's now a JBoost website
      - The website can be downloaded before destribution using the get_doc.sh
        script found in the /doc directory


+ Release 1.1 + 
===============

 - Added command line options 
      - Type of booster
      - Print usage help
      - Print version
 - Added two new boosting algorithms: BrownBoost and YabaBoost
 - The project is now on SourceForge

Known Bugs
 - There are still numerical instabilities in YabaBoost and BrownBoost.


+ Release 1.0 + 
===============

New Features and Enhancements
 - Cdded a configuration option booster_type to select which boosting 
   algorithm will be used. AdaBoost and LogLossBoost are avaiable.
 - Changed Booster into an interface and created AbstractBooster as a skeleton
   implementation
 - Added FileLoader to support loading data files by searching a collection of
   directories.
 - Created several JUnit tests for jboost.controller and jboost.booster packages.
   Also added an AllTests suite to run all the tests in the jboost project.
 - Added support for running tests from command-line ant. Most tests can also be
   run via JUnit support built into eclipse IDE.
 - Added junit.jar to /lib directory
 
Known Bugs
 - The FileLoader is not finding data files when using ant-based test target with
   the eclipse IDE.

 
+ Release 0.1 +
===============

Features
 - Can run the AdaBoost learning algorithm
 - Produces executable jar that can read and analyze data