Machine Learning Library?

Question

Which libraries do you use for machine learning for C/C++, Python, Java?

Answer 1

With Java, I find that Weka covers most of my needs.
I've had less luck with PyML under Python but that's probably more of a matter of my lack of familiarity with PyML than effective limitations or shortcomings of this library.

Answer 2

Weka (in Java) is definitely one of the most popular, and has been exposed to most other (including Python).

For C++, I recommend looking at Shark, but Shogun is also quite good. (see, also, this related SO question).

Answer 3

Weka is good. Rapid Miner is more comprehensive and more complicated.

Answer 4

Python related:

Orange: data mining framework
NLTK: natural language processing and text analytics
Gensim: Vector Space Modelling
Machine Learning: An Algorithmic Perspective: Although it's a book, we can touch the source code directly.
NumPy/SciPy related
Python related Machine Learning Library

If you really need a tons of Machine Learning packages for review. MayBe http://mloss.org/software/ is the place.

Answer 5

For Python:

• MDP is a signal processing library, and allows you to easily build pipelines of algorithms

• scikits.learn is a rapidly expanding collection of machine learning algorithm

Answer 6

For a standalone library, I highly recommend libsvm (available in C++, python, and other languages). It's a good, standard support vectorm machine library that supports a lot of kernels.