The main new feature of this release is a bridge to WEKA. It is now possible to use all algorithms implemented in WEKA from inside Java-ML, using the same interfaces.
This maintenance release of the Java Machine Learning Library adds a few minor features and fixes some bugs
* Support to load from zip and gzip
* Tanimoto is renamed to JaccardIndex to better reflect the true implementation
* added SpearmanFootrule
* fixed documentation in some places
* fixed bug in Logistic classifier
This release is a major upgrade from the previous version. Many optimizations and tweaks have been made to the core of the library. We have updated many of the existing algorithms with more documentation and a more logical structure.
The new version is available for download .
This release is mainly an enhancement to the documentation. We have also improved a lot on the build script and the release protocol so that much is done automagically and we can focus more on the actual development. We have also made the structure of the library more intuitive and added several tutorials.
This release contains numerous new algorithms. And some improvements in the current algorithms.
- We have include real wavelet transforms (Sombrero, Morlet, Gauss and Der Gauss).
- A fast implementation of the Dynamic Time Warping algorithm
- Fast Fourier Transform filter
- Support for Complex numbers
- Added discretization filters
- Added some more distance/similarity measure
- Added some feature selection techniques (PCA, FCBF)
The 11th installment of the Java Machine Learning Library has been released. The main features of this release are two new classification algorithms: a binary Support Vector Machine and a logistic regression classifier. Besides these two classifiers also numerous other improvements to the library were made.
The smaller improvements to the library include:
- restructuring of the library so that the structure is more logical
- added Linear and Polynomial kernels, two distance measures often found in the context of support vector machines
- added Iterable support on datasets
We have released version 0.0.10 of the Java Machine Learning Library.
* two filters
* library core restructuring
* tweaks to some clustering algorithms
* included the JAMA package
This release features two new Filters. First, the Remove filter that allows you to remove certain attributes from all the instances in a Dataset. And second, the Principle Components Analysis filter, which allows you to select only the most relevant attributes of a dataset.
This weekly release is only a minor improvement over the previous one. In the AQBC algorithm we fixed a bug that would make the algorithm return no clusters if there was at least one instance for which all attribute values are the same.
This release fixes bugs in clustering algorithms where some algorithms could return empty clusters in some cases.
Some minor improvements around the library: made some more methods to manipulate datasets and methods to compare arrays.
New implementation of the Adaptive Quality Based Clustering algorithm based on the original implementation in Matlab.
A (nearly) complete Ant script to make a release is now in the SVN repository.