Machine learning tools, platforms Summary

1. platforms and systems

• TensorFlow - TensorFlow is Google's second-generation machine learning system, built-in extended support deep learning of any calculation can be used to calculate the flow graphic expression, can be used TensorFlow
• PaddlePaddle - Baidu developed deep learning platform, with easy to use, efficient, flexible and scalable features, deep learning algorithms provide support for a number of products inside Baidu
• Apache SINGA - SINGA training is based on large data sets, distributed learning platform for large-scale conventional depth learning modules. SINGA support a variety of popular deep learning module
• Scikit Flow - to simplify the interface TensorFlow, Scikit imitation learning, and the user may be predictive analysis using data mining
• VELES - depth study of distributed application system, users only need to provide parameters, and the rest can be handed over VELES. VELES developed by Samsung is another TensorFlow
• SpeeDO - for the common hardware design of parallel depth learning system. SpeeDO does not require special I / O hardware, support for CPU / GPU clusters, can be easily deployed on a variety of cloud environments, such as AWS, Google GCE, Microsoft Azure, etc.

frame

• Torchnet - Facebook deep learning framework in order to accelerate research and open source AI
• LightGBM - an implementation of Microsoft's open source framework GBDT algorithms, support for parallel training with high efficiency. GBDT aims to solve the problem encountered in the massive data, so GBDT can better and faster for industrial practice
• Guagua - Hadoop iterative computational framework Guagua PayPal is a open-source machine learning framework Shifu subprojects, distributed mainly to solve the problem of model training
• Chainer - Chainer bridge the gap between theory algorithm depth study and practical application, deep learning and flexible framework
• Shifu - Fast and scalable machine learning framework for Hadoop
• KeystoneML - framework written in Scala, designed to simplify the construction of large-scale, end-to-machine learning pipeline, constructed based on Apache Spark
• LightNet - lightweight, versatile, full of deep learning framework based on Matlab. The purpose is to provide an easy to understand for the study and research depth, easy to use and efficient computing platform
• DeepLearningKit - depth learning open source framework for iOS, OS X, and the tvOS
• GoLearn - GoLearn is a machine learning framework implemented in Go
• YCML - machine learning framework written in Objective-C, and also supports Swift

Toolkits and libraries

• DMTK - Microsoft's open source distributed machine learning toolkit, including DMTK distributed machine learning framework for LightLDA trained topic model and distributed vector word
• CNTK - depth learning tools for the Microsoft open source speech recognition package, GPU capability by means of the very high efficiency toolkit
• DSSTNE - Amazon open source deep learning tool that can simultaneously support two graphics processors (GPU) involved in operation, mainly for intelligent search and recommendation
  Scikit-learn - Python machine learning project, simple and efficient algorithm library offers a range of the supervised learning and unsupervised learning algorithms for data mining and data analysis. SciKit-learn, covering almost all major machine learning algorithm
• Deeplearning4j - is written in Java and Scala's first commercial-grade open source distributed database depth study, the business environment is designed to plug and play as the goal, by default more, to avoid too much configuration, so that non-researchers It is also capable of rapid prototyping
• MXNet - lightweight and flexible and efficient learning library depth, allowing the use of alphanumeric and command programming programming
• CaffeOnSpark - Yahoo open source distributed deep learning package based on Hadoop / Spark's
• BigDL - Intel open source distributed database based on Apache Spark deep learning, support for high-performance big data analytics
• Swift AI - high performance artificial intelligence and machine learning library written entirely in Swift, currently supports iOS and OS X, including a set of common tools of artificial intelligence and machine learning
• Gorgonia - Go machine learning library for writing mathematical formulas and evaluate multi-dimensional arrays. Theano and TensorFlow with similar ideas, support for GPU / CUDA, support distributed computing
• Shark C ++ - fast, modular, feature-rich open source machine learning C ++ library that provides a variety of machine learning technologies, such as linear / non-linear optimization, kernel-based learning algorithms, neural networks, etc.
• MLPACK - C ++ library of machine learning, highlights in its scalability, high speed and ease of use. Designed to allow new users to learn to use the machine through a simple, consistent API, while providing maximum flexibility and high performance of C ++ for professional users
• smile - Java library contains a variety of existing machine learning algorithms. E.g. adjacency matrix table and graph algorithms, based visualization Swing libraries
• PredictionIO - open source server machine learning, development engineers and data analysts can use it to build smart application, you can also do some forecasting features, such as personalized recommendations, and other content found
• Aerosolve - Airbnb pricing support the proposed system of machine learning engine
• Vowpal Wabbit - machine learning system that uses such as online, hash, reduce, reduce, learn, search, active and interactive learning technology to promote the development of cutting-edge machine learning techniques
• Apache SystemML - SystemML is flexible, scalable machine learning (ML) language, written in Java. It provides automatic optimization function, by clustering feature data and ensure efficient and scalable. SystemML runs MapReduce environment or Spark