Java Machine Learning Library 0.1.0 released

Version 0.1.0 is a major upgrade from the previous release (0.0.16). The core of the library has been rewritten entirely and this release is as such no longer compatible with previous releases.

API documentation

Main features:

  • The library now supports data sparsity. Instances now extend the Map<Integer,Double> interfaces, where the key is an attribute index and the value is the value for that attribute.
  • Dataset now extends the List<Instance> interface making all Collections operations available on data sets
  • Dataset and Instance are now fully mutable, which improves the performance of many algorithms significantly as it is no longer required to copy Instances all the time. The drawback however is that algorithm can modify a data set so you will have to reload your data sometimes after training.
  • Due to this core redesign some algorithms are temporarily unavailable but they will be re-instituted in the following releases. If you have things you need back right away, contact us.
  • A number of new algorithms have been added.
  • If you require assistance porting to the new interfaces, please contact us, we will help where possible.

    In other news, we launch the new web site for Java-ML together with this release. This site offers a starting point for Java-ML users and developers. The old site, located at will be phased out in the coming months and all information will be migrated to the new site which is located at