Description:

Debellor is a scalable and extensible platform which provides common architecture for data mining and machine learning algorithms of various types. The algorithms can be combined together to build data processing schemes of large complexity. The unique feature of Debellor is data streaming, which enables efficient processing of large volumes of data.

Debellor simplifies implementation of composite data mining algorithms:

• Over 120 algorithms are already available and can be used as building blocks. These include all classifiers from Weka and Rseslib libraries, all filters from Weka and a reader of arff files.
• All algorithms are accessible through the same simple interface of a Cell.
• Thanks to stream architecture, algorithms may process data on the fly, without buffering in memory. This enables efficient handling of large volumes of data and gives freedom of designing arbitrarily complex algorithms.
• New data types may be defined, specific to a particular application domain.
• Multi-threading: composite experiments may be executed concurrently in many threads. Debellor takes care of thread management and synchronization, the user does not have to know anything about concurrency.

Changes to previous version:

• Naming of numerous classes/methods/fields changed to be more accurate and comprehensible
• Weka and Rseslib libraries updated to the newest versions: Weka 3.6.1 & Rseslib 3.0.1. Debellor's wrappers adapted
• New class: CrossValidation - evaluator of trainable cells through cross-validation
• New class: RMSE - calculation of Root Mean Squared Error score
• Data objects can be compared and used in collections
• ArffReader can read from a user-provided java.io.InputStream
• More convenient use of parameters (setting values)
• More convenient use of data objects and data types (construction, type casting)
• Other minor improvements to existing classes
• Javadoc extended

BibTeX Entry: Download

Corresponding Paper BibTeX Entry: Download

Supported Operating Systems: Agnostic

Data Formats: Arff

Tags: Online Learning, Machine Learning, Framework, Data Mining, Nips2008

Archive: download here

Other available revisons

Version	Changelog	Date
1.0	Naming of numerous classes/methods/fields changed to be more accurate and comprehensible Weka and Rseslib libraries updated to the newest versions: Weka 3.6.1 & Rseslib 3.0.1. Debellor's wrappers adapted New class: CrossValidation - evaluator of trainable cells through cross-validation New class: RMSE - calculation of Root Mean Squared Error score Data objects can be compared and used in collections ArffReader can read from a user-provided java.io.InputStream More convenient use of parameters (setting values) More convenient use of data objects and data types (construction, type casting) Other minor improvements to existing classes Javadoc extended	July 30, 2009, 16:48:05
0.6.1	Initial Announcement on mloss.org.	September 27, 2008, 22:11:28

Comments

No one has posted any comments yet. Perhaps you'd like to be the first?

Leave a comment

You must be logged in to post comments.