17 Great Machine Learning Libraries

17 Great Machine Learning Libraries
Java

• Spark: Apache’s new upstart, supposedly up to a hundred times faster than Hadoop, now includes MLLib, which contains a good selection of machine learning algorithms, including classification, clustering and recommendation generation. Currently undergoing rapid development. Development can be in Python as well as JVM languages.
• Mahout: Apache’s machine learning framework built on top of Hadoop, this looks promising, but comes with all the baggage and overhead of Hadoop.
• Weka: this is a Java based library with a graphical user interface that allows you to run experiments on small datasets. This is great if you restrict yourself to playing around to get a feel for what is possible with machine learning. However, I would avoid using this in production code at all costs: the API is very poorly designed, the algorithms are not optimised for production use and the documentation is often lacking.
• Mallet: another Java based library with an emphasis on document classification. I’m not so familiar with this one, but if you have to use Java this is bound to be better than Weka.
• JSAT: stands for “Java Statistical Analysis Tool” - created by Edward Raff and was born out of his frustation with Weka (I know the feeling). Looks pretty cool.

• LibSVM and LibLinear: these are C libraries for support vector machines; there are also bindings or implementations for many other languages. These are the libraries used for support vector machine learning in Scikit-learn.

Read full article from 17 Great Machine Learning Libraries