deep-learning-frameworks

From: http://venturebeat.com/2015/11/14/deep-learning-frameworks/

Here’s a rundown of some other notable deep learning libraries available today.

The big ones

• Caffe. With roots at the University of California, Berkeley, Caffe is widely used, including at Pinterest. Based in C++, Caffe was the basis of the DeepDream project that Google released earlier this year to show how its artificial neural networks viewed images.
• Theano. No, it’s not the ancient Greek philosopher. It’s a Python packagethat has spawned a slew of other Python packages for deep learning, including, most prominently, Blocks, Keras, Lasagne, and OpenDeep. Theano originated in 2008 at the Montreal Institute for Learning Algorithms, a division of the University of Montreal and the home base of deep learning luminary Yoshua Bengio.
• Torch. This framework has been around for more than a decade, but in the past year and a half, Facebook has demonstrated a major commitment to Torch, by open-sourcing deep learning modules and extensions for it. Twitter is on board, too. One quirk of Torch is its dependency on the programming language Lua, which is not especially popular these days but does have a record of being used to build video games.

The long tail

• Apache Singa. Written almost completely in C++, this one emerged as an Apache incubator project in March. The original work on Singa was done by six students and research fellows from the National University of Singapore and one professor at China’s Zhejiang University.
• Brainstorm. A promising new library from a small team of researchers at the Swiss artificial intelligence lab Istituto Dalle Molle di Studi sull’Intelligenza Artificiale (IDSIA), Brainstorm can handle what are being called Highway Networks involving very deep networks with hundreds of layers. Like Theano, it’s written in Python.
• Chainer. Preferred Networks, a startup based in Japan, announced the release of the Python-based framework in June. Chainer’s design is based on the principle “define by run” — that is, the network is determined on the fly rather than only at the beginning, according to the documentation for the framework.
• ConvNetJS. This tool from Stanford University Ph.D. student Andrej Karpathy allows you to train neural nets right inside of your browser, using good old JavaScript. Karpathy has a great tutorial for getting started with ConvNetJS, as well as nifty browser-based demos.
• Deeplearning4j. The name says it all — it’s deep learning for Java. This project is backed by startup Skymind, which launched in June 2014. Users of the software include Accenture, Booz Allen, Chevron, and IBM.
• h2o. This Java-based framework is part of a more general machine learning runtime from a startup that goes by the same name (although not long agothe startup went by a different name, 0xdata).
• Marvin. This new entrant from Princeton University’s Vision Group is written in C++. The team offers a file for converting Caffe models into a format that works in Marvin.
• MatConvNet. This is a MATLAB toolbox for implementing convolutional neural nets. It was first developed by professor Andrea Vedaldi and Ph.D. student Karel Lenc of the University of Oxford’s Robotics Research Group.
• MXNet. Primarily written in C++, MXNet was created by the people behind the CXXNet, Minerva, and Purine2 projects. It’s meant to use memory efficiently, and can even run on a smartphone, for tasks like image recognition.
• Neon. Startup Nervana Systems published its Neon software under an open source back in May. Some benchmarks suggest that Neon — which is written mostly in Python and Sass — could outperform Caffe, Torch, and Google’s TensorFlow.
• Veles. Named after the Slavic god by the same name, Veles comes fromSamsung. It’s written mostly in Python, and it can be run in an IPython notebook.