Context

#

Learned in this study

#

Things to explore

#

Overview

#

Accord.NET framework

#

• Website: http://accord-framework.net/
• Language: C#/.NET
• Type: Machine learning framework

Blocks

#

• Website: https://github.com/mila-udem/blocks
• Language: Python
• Type: Neural network framework

Details

#

• Built on Theano
• Specialized for neural network models

Caffe

#

• Website: http://caffe.berkeleyvision.org/
• Language: C++
• Type: Neural network framework

Chainer

#

• Website: http://chainer.org/
• Language: Python
• Type: Neural network framework

Deeplearning4j

#

• Website: http://deeplearning4j.org/
• Language: Java/Scala
• Type: Neural network framework

GoLearn

#

• Website: https://github.com/sjwhitworth/golearn
• Language: Go
• Type: Neural network framework

Keras

#

• Website: http://keras.io/
• Language: Python
• Type: Neural network framework

Details

#

• Built on Theano/Tensorflow as backend
• The goal is to make it easy/fast to prototype

Lasagne

#

• Website: https://github.com/Lasagne/Lasagne
• Language: Python
• Type: Neural network framework

Details

#

• Built on Theano

MLlib

#

• Website: https://spark.apache.org/mllib/
• Language: Java/Python
• Type: Machine learning framework

mlpack

#

• Website: http://www.mlpack.org/
• Language: C++
• Type: Machine learning framework

NuPIC

#

• Website: http://numenta.org/
• Language: Python
• Type: Machine learning framework

Details

#

• Hierarchical Temporal Memory only

Oryx

#

• Website: http://oryx.io/
• Language: Java
• Type: Machine learning framework

Details

#

• Built on Apache Spark and Apache Kafka

Pybrain

#

• Website: http://pybrain.org/
• Language: Python
• Type: Machine learning framework

Details

#

• Small documentation

Pylearn2

#

• Website: https://github.com/lisa-lab/pylearn2
• Language: Python
• Type: Machine learning framework

Details

#

• Built on Theano
• No more maintainer

Scikit-learn

#

• Website: http://scikit-learn.org/
• Language: Python
• Type: Machine learning framework

Details

#

• Excellent documentation

Tensorflow

#

• Website: https://www.tensorflow.org/
• Language: Python
• Used by: Google

Details

#

• Supports only Mac OS X and Ubuntu/Linux 64-bit
• GPU support
• No distributed architecture yet

Theano

#

• Website: http://deeplearning.net/software/theano/
• Language: Python
• Type: Scientific computing framework
• Used by: MILA

Details

#

• Supports all platforms, however mainly developed/tested on Linux 64-bit
• GPU support
• No distributed architecture yet

Torch

#

• Website: http://torch.ch/
• Language: LUA
• Type: Scientific computing framework
• Used by: Facebook, Google, Twitter, NYU

Details

#

• Supports only Mac OS X and Ubuntu
• Community managed documentation (discutable quality)
• GPU support

See also

#

References

#

• https://github.com/josephmisiti/awesome-machine-learning
• http://docs.chainer.org/en/stable/comparison.html
• https://github.com/szilard/benchm-ml
• https://github.com/szilard/survey-ml-tools
• http://scikit-learn.org/ml-benchmarks/