![\"machine-learning-framework\"](\"https:\/\/i0.wp.com\/opensourceforu.com\/wp-content\/uploads\/2016\/12\/Machine-Learning-Framework-350x307.jpg?resize=640%2C561\")
<\/figure>\n\n\n\nIn this article, we present what the author rates as the top eight open source machine learning frameworks.<\/em><\/p>\n\n\n\n Learning may be defined as the process of improving one\u2019s ability to \nperform a task efficiently. Machine learning is another sub-field of \ncomputer science, which enables modern computers to learn without being \nexplicitly programmed. Machine learning has basically evolved from \nartificial intelligence via pattern recognition and computational \nlearning theory. Machine learning explores the area of algorithms, which\n can make high end predictions on data. In recent times, machine \nlearning has been deployed in a wide range of computing tasks, where \ndesigning efficient algorithms and programs becomes rather difficult, \nsuch as email spam filtering, optical character recognition, search \nengine improvement, digital image processing, data mining, etc. The application of machine learning to diverse areas of computing is \ngaining popularity rapidly, not only because of cheap and powerful \nhardware, but also because of the increasing availability of free and \nopen source software, which enable machine learning to be implemented \neasily. Machine learning practitioners and researchers, being a part of \nthe software engineering team, continuously build sophisticated \nproducts, integrating intelligent algorithms with the final product to \nmake software work more reliably, quickly and without hassles. Apache Singa<\/strong> Its main features are:<\/p>\n\n\n\n The latest version is 1.0. Shogun<\/strong> Apache Mahout<\/strong> Apache Spark MLlib<\/strong> TensorFlow<\/strong> Oryx 2<\/strong> Oryx 2 consists of the following layers of Lambda architecture as well as connecting elements.<\/p>\n\n\n\n Accord.NET<\/strong><\/p>\n\n\n\n Accord.NET is a .NET open source machine learning framework for \nscientific computing, and consists of multiple libraries for diverse \napplications like statistical data processing, pattern recognition, \nlinear algebra, artificial neural networks, image and signal processing,\n etc. The latest version is 3.1.0. Amazon Machine Learning (AML)<\/strong><\/p>\n\n\n\n Amazon Machine Learning (AML) is a machine learning service for \ndevelopers. It has many visualisation tools and wizards for creating \nhigh-end sophisticated and intelligent machine learning models without \nany need to learn complex ML algorithms and technologies. Via AML, \npredictions for applications can be obtained using simple APIs without \nusing custom prediction generation code or complex infrastructure.<\/p>\n\n\n\n AML is based on simple, scalable, dynamic and flexible ML technology \nused by Amazon\u2019s \u2018Internal Scientists\u2019 community professionals to create\n Amazon Cloud Services. AML connects to data stored in Amazon S3, \nRedshift or RDS, and can run binary classification, multi-class \ncategorisation or regression on this data to create models. Its key features are:<\/p>\n\n\n\n Website:<\/strong> https:\/\/aws.amazon.com\/machine-learning\/ <\/em><\/a><\/p>\n\n\n\n
\nTom M. Mitchell, renowned computer scientist and professor at Carnegie \nMellon University, USA, defined machine learning as: \u201cA computer program\n is said to learn from experience E with respect to some class of tasks T\n and performance measure P, if its performance at tasks in T, as \nmeasured by P, improves with experience E.\u201d
\nMachine learning tasks are broadly classified into three categories, \ndepending on the nature of the learning \u2018signal\u2019 or \u2018feedback\u2019 available\n to a learning system.<\/p>\n\n\n\n
\nThere is a wide range of open source machine learning frameworks \navailable in the market, which enable machine learning engineers to \nbuild, implement and maintain machine learning systems, generate new \nprojects and create new impactful machine learning systems.
\nLet\u2019s take a look at some of the top open source machine learning frameworks available.\n\nAdvertisement\n\n\n\n\n\n\n\n\n<\/p>\n\n\n\n
\nThe Singa Project was initiated by the DB System Group at the National \nUniversity of Singapore in 2014, with a primary focus on distributed \ndeep learning by partitioning the model and data onto nodes in a cluster\n and parallelising the training. Apache Singa provides a simple \nprogramming model and works across a cluster of machines. It is \nprimarily used in natural language processing (NLP) and image \nrecognition. A Singa prototype accepted by Apache Incubator in March \n2015 provides a flexible architecture of scalable distributed training \nand is extendable to run over a wide range of hardware.
\nApache Singa was designed with an intuitive programming model based on \nlayer abstraction. A wide variety of popular deep learning models are \nsupported, such as feed-forward models like convolutional neural \nnetworks (CNN), energy models like Restricted Boltzmann Machine (RBM), \nand recurrent neural networks (RNN). Based on a flexible architecture, \nSinga runs various synchronous, asynchronous and hybrid training \nframeworks.
\nSinga\u2019s software stack has three main components: Core, IO and Model. \nThe Core component is concerned with memory management and tensor \noperations. IO contains classes for reading and writing data to the disk\n and the network. Model includes data structures and algorithms for \nmachine learning models.<\/p>\n\n\n\n
\nWebsite:<\/strong> http:\/\/singa.apache.org\/en\/index.html<\/em><\/a><\/p>\n\n\n\n
\nShogun was initiated by Soeren Sonnenburg and Gunnar Raetsch in 1999 and\n is currently under rapid development by a large team of programmers. \nThis free and open source toolbox written in C++ provides algorithms and\n data structures for machine learning problems. Shogun Toolbox provides \nthe use of a toolbox via a unified interface from C++, Python, Octave, \nR, Java, Lua and C++; and can run on Windows, Linux and even MacOS. \nShogun is designed for unified large-scale learning for a broad range of\n feature types and learning settings, like classification, regression, \ndimensionality reduction, clustering, etc. It contains a number of \nexclusive state-of-art algorithms, such as a wealth of efficient SVM \nimplementations, multiple kernel learning, kernel hypothesis testing, \nKrylov methods, etc.
\nShogun supports bindings to other machine learning libraries like \nLibSVM, LibLinear, SVMLight, LibOCAS, libqp, VowpalWabbit, Tapkee, SLEP,\n GPML and many more.
\nIts features include one-time classification, multi-class \nclassification, regression, structured output learning, pre-processing, \nbuilt-in model selection strategies, visualisation and test frameworks; \nand semi-supervised, multi-task and large scale learning.
\nThe latest version is 4.1.0.
\nWebsite:<\/strong> http:\/\/www.shogun-toolbox.org\/<\/em><\/a><\/p>\n\n\n\n
\nApache Mahout, being a free and open source project of the Apache \nSoftware Foundation, has a goal to develop free distributed or scalable \nmachine learning algorithms for diverse areas like collaborative \nfiltering, clustering and classification. Mahout provides Java libraries\n and Java collections for various kinds of mathematical operations.
\nApache Mahout is implemented on top of Apache Hadoop using the MapReduce\n paradigm. Once Big Data is stored on the Hadoop Distributed File System\n (HDFS), Mahout provides the data science tools to automatically find \nmeaningful patterns in these Big Data sets, turning this into \u2018big \ninformation\u2019 quickly and easily.<\/p>\n\n\n\n
\nThe latest version is 0.12.2.
\nWebsite:<\/strong> https:\/\/mahout.apache.org\/<\/em><\/a><\/li><\/ul>\n\n\n\n
\nApache Spark MLlib is a machine learning library, the primary objective \nof which is to make practical machine learning scalable and easy. It \ncomprises common learning algorithms and utilities, including \nclassification, regression, clustering, collaborative filtering, \ndimensionality reduction as well as lower-level optimisation primitives \nand higher-level pipeline APIs.
\nSpark MLlib is regarded as a distributed machine learning framework on \ntop of the Spark Core which, mainly due to the distributed memory-based \nSpark architecture, is almost nine times as fast as the disk-based \nimplementation used by Apache Mahout.
\nThe various common machine learning and statistical algorithms that have been implemented and included with MLlib are:<\/p>\n\n\n\n
\nThe latest version is 2.0.1.
\nWebsite:<\/strong> http:\/\/spark.apache.org\/mllib\/ <\/em><\/a><\/li><\/ul>\n\n\n\n
\nTensorFlow is an open source software library for machine learning \ndeveloped by the Google Brain Team for various sorts of perceptual and \nlanguage understanding tasks, and to conduct sophisticated research on \nmachine learning and deep neural networks. It is Google Brain\u2019s second \ngeneration machine learning system and can run on multiple CPUs and \nGPUs. TensorFlow is deployed in various products of Google like speech \nrecognition, Gmail, Google Photos and even Search.
\nTensorFlow performs numerical computations using data flow graphs. These\n elaborate the mathematical computations with a directed graph of nodes \nand edges. Nodes implement mathematical operations and can also \nrepresent endpoints to feed in data, push out results or read\/write \npersistent variables. Edges<\/em> describe the input\/output relationships between nodes. Data edges<\/em> carry dynamically-sized multi-dimensional data arrays or tensors.
\nIts features are listed below.<\/p>\n\n\n\n
\nThe latest version is 0.10.0.
\nWebsite:<\/strong> www.tensorflow.org <\/em><\/li><\/ul>\n\n\n\n
\nOryx 2 is a realisation of Lambda architecture built on Apache Spark and\n Apache Kafka for real-time large scale machine learning. It is designed\n for building applications and includes packaged, end-to-end \napplications for collaborative filtering, classification, regression and\n clustering.
\nOryx 2 comprises the following three tiers.<\/p>\n\n\n\n
\nThe latest version is 2.2.1.
\nWebsite:<\/strong> http:\/\/oryx.io\/<\/em><\/a><\/li><\/ul>\n\n\n\n
\nThe framework is divided into libraries via the installer, compressed \narchives and NuGet packages, which include Accord.Math, \nAccord.Statistics, Accord.MachineLearning, Accord.Neuro, Accord.Imaging,\n Accord.Audio, Accord.Vision, Accord.Controls, Accord.Controls.Imaging, \nAccord.Controls.Audio, Accord.Controls.Vision, etc.
\nIts features are:<\/p>\n\n\n\n
\nWebsite:<\/strong> www.accord-framework.net<\/em><\/a><\/p>\n\n\n\n
\nThe key contents used in Amazon ML are listed below.<\/p>\n\n\n\n