Spark is the Future of Analytics

At the 2016 Spark Summit, Gartner Research Director Nick Heudecker asked: Is Spark the Future of Data Analysis?  It’s an interesting question, and it requires a little parsing. Nobody believes that Spark alone is the future of data analysis, even its most ardent proponents. A better way to frame the question: Does Spark have a role in the future of analytics? What is that role?

Unfortunately, Heudecker didn’t address the question but spent the hour throwing shade at Spark.

Spark is overhyped! He declared. His evidence? This:


One might question an analysis that equates real things like optimization with fake things like “Citizen Data Science.” Gartner’s Hype Cycle by itself proves nothing; it’s a conceptual salad, with neither empirical foundation nor predictive power.

If you want to argue that Spark is overhyped, produce some false or misleading claims by project principals, or documented cases where the software failed to work as claimed. It’s possible that such cases exist. Personally, I don’t know of any, and neither does Nick Heudecker, or he would have included them in his presentation.

Instead, he cited a Gartner survey showing that organizations don’t use Spark and Flink as much as they use other tools for data analysis. From my notes, here are the percentages:

  • EDW: 57%
  • Cloud: 44%
  • Hadoop: 42%
  • Stat Packages: 32%
  • Spark or Flink: 9%
  • Graph Databases: 8%

That 42% figure for Hadoop is interesting. In 2015, Gartner concern-trolled the tech community, trumpeting the finding that “only” 26% of respondents in a survey said they were “deploying, piloting or experimenting with Hadoop.” So — either Hadoop adoption grew from 26% to 42% in a year, or Gartner doesn’t know how to do surveys.

In any event, it’s irrelevant; statistical packages have been available for 40 years, EDWs for 25, Spark for 3. The current rate of adoption for a project in its youth tells you very little about its future. It’s like arguing that a toddler is cognitively challenged because she can’t do integral calculus without checking the Wolfram app on her iPad.

Heudecker closed his presentation with the pronouncement that he had no idea whether or not Spark is the future of data analysis, and bolted the venue faster than a jackrabbit on Ecstasy. Which begs the question: why pay big bucks for analysts who have no opinion about one of the most active projects in the Big Data ecosystem?

Here are eight reasons why Spark has a central role in the future of analytics.

(1) Nearly everyone who uses Hadoop will use Spark.

If you believe that 42% of enterprises use Hadoop, you must believe that 41.9% will use Spark. Every Hadoop distribution includes Spark. Hive and Pig run on Spark. Hadoop early adopters will gradually replace existing MapReduce applications and build most new applications in Spark. Late adopters may never use MapReduce.

The only holdouts for MapReduce will be those who want their analysis the way they want their barbecue: low and slow.

Of course, Hadoop adoption isn’t static. Forrester’s Mike Gualtieri argues that 100% of enterprises will use Hadoop within a few years.

(2) Lots of people who don’t use Hadoop will use Spark.

For Hadoop users, Spark is a fast replacement for MapReduce. But that’s not all it is. Spark is also a general-purpose data processing environment for advanced analytics. Hadoop has baggage that data science teams don’t need, so it’s no surprise to see that most Spark users aren’t using it with Hadoop. One of the key advantages of Spark is that users aren’t tied to a particular storage back end, but can choose from many different options. That’s essential in real-world data science.

(3) For scalable open source data science, Spark is the only game in town.

If you want to argue that Spark has no future, you’re going to have to name an alternative. I’ll give you a minute to think of something.

Time’s up.

You could try to approximate Spark’s capabilities with a collection of other projects: for example, you could use Presto for SQL, H2O for machine learning, Storm for streaming, and Giraph for graph analysis. Good luck pulling those together. was one of the first vendors to build an interface to Spark because even if you want to use H2O for machine learning, you’re still going to use Spark for data wrangling.

“What about Flink?” you ask. Well, what about it? Flink may have a future, too, if anyone ever supports it other than ten guys in a loft on the Tempelhofer Ufer. Flink’s event-based runtime seems well-suited for “pure” streaming applications, but that’s low-value bottom-of-the-stack stuff. Flink’s ML library is still pretty limited, and improving it doesn’t appear to be a high priority for the Flink team.

(4) Data scientists who work exclusively with “small data” still need Spark.

Data scientists satisfy most business requests for insight with small datasets that can fit into memory on a single machine. Even if you measure your largest dataset in gigabytes, however, there are two ways you need Spark: to create your analysis dataset and to parallelize operations.

Your analysis dataset may be small, but it comes from a larger pool of enterprise data. Unless you have servants to pull data for you, at some point you’re going to have to get your hands dirty and deal with data at enterprise scale. If you are lucky, your organization has nice clean data in a well-organized data warehouse that has everything anyone will ever need in a single source of truth.

Ha ha! Just kidding. Single sources of truth don’t exist, except in the wildest fantasies of data warehouse vendors. In reality, you’re going to muck around with many different sources and integrate your analysis data on the fly. Spark excels at that.

For best results, machine learning projects require hundreds of experiments to identify the best algorithm and optimal parameters. If you run those tests serially, it will take forever; distribute them across a Spark cluster, and you can radically reduce the time needed to find that optimal model.

(5) The Spark team isn’t resting on its laurels.

Over time, Spark has evolved from a research project for scalable machine learning to a general purpose data processing framework. Driven by user feedback, Spark has added SQL and streaming capabilities, introduced Python and R APIs, re-engineered the machine learning libraries, and many other enhancements.

Here are some projects under way to improve Spark:

— Project Tungsten, an ongoing effort to optimize CPU and memory utilization.

— A stable serialization format (possibly Apache Arrow) for external code integration.

— Integration with deep learning frameworks, including TensorFlow and Intel’s new BigDL library.

— A cost-based optimizer for Spark SQL.

— Improved interfaces to data sources.

— Continuing improvements to the Python and R APIs.

Performance improvement is an ongoing mission; for selected operations, Spark 2.0 runs 10X faster than Spark 1.6.

(6) More cool stuff is on the way.

Berkeley’s AMPLab, the source of Spark, Mesos, and Tachyon/Alluxio, is now RISELab. There are four projects under way at RISELab that will extend Spark capabilities:

Clipper is a prediction serving system that brokers between machine learning frameworks and end-user applications. The first Alpha release, planned for mid-April 2017, will serve scikit-learn, Spark ML and Spark MLLib models, and arbitrary Python functions.

Drizzle, an execution engine for Apache Spark, uses group scheduling to reduce latency in streaming and iterative operations. Lead developer Shivaram Venkataraman has filed a design document to implement this approach in Spark.

Opaque is a package for Spark SQL that uses Intel SGX trusted hardware to deliver strong security for DataFrames. The project seeks to enable analytics on sensitive data in an untrusted cloud, with data encryption and access pattern hiding.

Ray is a distributed execution engine for Spark designed for reinforcement learning.

Three Apache projects in the Incubator build on Spark:

— Apache Hivemall is a scalable machine learning library implemented as a collection of Hive UDFs designed to run on Hive, Pig or Spark SQL with MapReduce, Tez or Spark.

— Apache PredictionIO is a machine learning server built on top of an open source stack, including Spark, HBase, Spray, and Elasticsearch.

— Apache SystemML is a library of machine learning algorithms that run on Spark and MapReduce, originally developed by IBM Research.

MIT’s CSAIL lab is working on ModelDB, a system to manage machine learning models. ModelDB extracts and stores model artifacts and metadata, and makes this data available for easy querying and visualization. The current release supports Spark ML and scikit-learn.

(7) Commercial vendors are building on top of Spark.

The future of analytics is a hybrid stack, with open source at the bottom and commercial software for business users at the top. Here is a small sample of vendors who are building easy-to-use interfaces atop Spark.

Alpine Data provides a collaboration environment for data science and machine learning that runs on Spark (and other platforms.)

AtScale, an OLAP on Big Data solution, leverages Spark SQL and other SQL engines, including Hive, Impala, and Presto.

Dataiku markets Data Science Studio, a drag-and-drop data science workflow tool with connectors for many different storage platforms, scikit-learn, Spark ML and XGboost.

StreamAnalytix, a drag-and-drop platform for real-time analytics, supports Spark SQL and Spark Streaming, Apache Storm, and many different data sources and sinks.

Zoomdata, an early adopter of Spark, offers an agile visualization tool that works with Spark Streaming and many other platforms.

All of the leading agile BI tools, including Tableau, Qlik, and PowerBI, support Spark. Even stodgy old Oracle’s Big Data Discovery tool runs on Spark in Oracle Cloud.

(8) All of the leading commercial advanced analytics platforms use Spark.

All of them, including SAS, a company that embraces open source the way Sylvester the Cat embraces a skunk. SAS supports Spark in SAS Data Loader for Hadoop, one of SAS’ five different Hadoop architectures. (If you don’t like SAS architecture, wait six months for another.)

Magic Quadrant for Advanced Analytics Platforms, 2016

— IBM embraces Spark like Romeo embraced Juliet, hopefully with a better ending. IBM contributes heavily to the Spark project and has rebuilt many of its software products and cloud services to use Spark.

— KNIME’s Spark Executor enables users of the KNIME Analytics Platform to create and execute Spark applications. Through a combination of visual programming and scripting, users can leverage Spark to access data sources, blend data, train predictive models, score new data, and embed Spark applications in a KNIME workflow.

— RapidMiner’s Radoop module supports visual programming across SparkR, PySpark, Pig, and HiveQL, and machine learning with SparkML and H2O.

— Statistica, which is no longer part of Dell, offers Spark integration in its Expert and Enterprise editions.

— Microsoft supports Spark in AzureHD, and it has rebuilt Microsoft R Server’s Hadoop integration to leverage Spark as well as MapReduce. VentureBeat reports that Databricks will offer its managed service for Spark on Microsoft Azure later this year.

— SAP, another early adopter of Spark, supports Vora, a connector to SAP HANA.

You get the idea. Spark is deeply embedded in the ecosystem, and it’s foolish to argue that it doesn’t play a central role in the future of analytics.

The Year in Machine Learning (Part Two)

This is the second installment in a four-part review of 2016 in machine learning and deep learning. Part One, here, covered general trends. In Part Two, we review the year in open source machine learning and deep learning projects. Parts Three and Four will cover commercial machine learning and deep learning software and services.

There are thousands of open source projects on the market today, and we cannot cover them all. We’ve selected the most relevant projects based on usage reported in surveys of data scientists, as well as development activity recorded in OpenHub.  In this post, we limit the scope to projects with a non-profit governance structure, and those offered by commercial ventures that do not also provide licensed software. Part Three will include software vendors who offer open source “community” editions together with commercially licensed software.

R and Python maintained their leadership as primary tools for open data science. The Python versus R debate continued amid an emerging consensus that data scientists should consider learning both. R has a stronger library of statistics and machine learning techniques and is agiler when working with small data. Python is better suited to developing applications, and the Python open source license is less restrictive for commercial application development.

Not surprisingly, deep learning frameworks were the most dynamic category, with TensorFlow, Microsoft Cognitive, and MXNet taking leadership away from more mature tools like Caffe and Torch. It’s remarkable that deep learning tools introduced as recently as 2014 now seem long in the tooth.

The R Project

The R user community continued to expand in 2016. It ranked second only to SQL in the 2016 O’Reilly Data Science Salary Survey; first in the KDNuggets poll; and first in the Rexer survey. R ranked fifth in the IEEE Spectrum ranking.

R functionality grew at a rapid pace. In April, Microsoft’s Andrie de Vries reported that there were more than 8,000 packages in CRAN, R’s primary repository for contributed packages. As of mid-December, there are 9,737 packages.  Machine learning packages in CRAN continued to grow in number and functionality.

The R Consortium, a Collaborative Project of the Linux Foundation, made some progress in 2016. IBM and ESRI joined the Consortium, whose membership now also includes Alteryx, Avant, DataCamp, Google, Ketchum Trading, Mango Solutions, Microsoft, Oracle, RStudio, and TIBCO. There are now three working groups and eight funded projects.

Hadley Wickham had a good year. One of the top contributors to the R project, Wickham co-wrote R for Data Science and released tidyverse 1.0.0 in September. In The tidy tools manifesto, Wickham explained the four basic principles to a tidy API.

Max Kuhn, the author of Applied Predictive Modeling and developer of the caret package for machine learning, joined RStudio in November. RStudio previously hired Joseph Rickert away from Microsoft.

AT&T Labs is doing some impressive work with R, including the development of a distributed back-end for out-of-core processing with Hadoop and other data platforms. At the UseR! Conference, Simon Urbanek presented a summary.

It is impossible to enumerate all of the interesting analysis performed in R this year. David Robinson’s analysis of Donald Trump’s tweets resonated; using tidyverse, tidytext, and twitteR, Robinson was able to distinguish between the candidate’s “voice” and that of his staffers on the same account.

On the Revolutions blog, Microsoft’s David Smith surveyed the growing role of women in the R community.

Microsoft and Oracle continued to support enhanced R distributions; we’ll cover these in Part Three of this survey.


Among data scientists surveyed in the 2016 KDNuggets poll, 46% said they use Python for analytics, data mining, data science or machine learning projects in the past twelve months. That figure was up from 30% in 2015, and second only to R. In the 2016 O’Reilly Data Science Salary Survey, Python ranked third behind SQL and R.

Python Software Foundation (PSF) expanded the number and dollar value of its grants. PSF awarded many small grants to groups around the world that promote Python education and training. Other larger grants went to projects such as the design of the Python in Education site, improvements to the packaging ecosystem (see below), support for the Python 3.6 beta 1 release sprint, and support for major Python conferences.

The Python Packaging Authority launched the Warehouse project to replace the existing Python Packaging Index (PyPI.) Goals of the project include updating the visual identity, making packages more discoverable and improving support for package users and maintainers.

PSF released Python 3.6.0 and Python 2.7.13 in December.  The scikit-learn team released Version 0.18 with many enhancements and bug fixes; maintenance release Version 0.18.1 followed soon after that.

Many of the key developments for machine learning in Python were in the form of Python APIs to external packages, such as Spark, TensorFlow, H2O, and Theano. We cover these separately below.

Continuum Analytics expanded its commercial support for Python during the year and added commercially licensed software extensions which we will cover in Part Three.

Apache Software Foundation

There are ten Apache projects with machine learning capabilities. Of these, Spark has the most users, active contributors, commits, and lines of code added. Flink is a close second in active development, although most Flink devotees care more about its event-based streaming than its machine learning capabilities.

Top-Level Projects

There are four top-level Apache projects with machine learning functionality: Spark, Flink, Mahout, and OpenNLP.

Apache Spark

The Spark team delivered Spark 2.0, a major release, and six maintenance releases. Key enhancements to Spark’s machine learning capabilities in this release included additional algorithms in the DataFrames-based API, in PySpark and in SparkR, as well as support for saving and loading ML models and pipelines. The DataFrames-based API is now the primary interface for machine learning in Spark, although the team will continue to support the RDD-based API.

GraphX, Spark’s graph engine, remained static. Spark 2.0 included many other enhancements to Spark’s SQL and Streaming capabilities.

Third parties added 24 machine learning packages to Spark Packages in 2016.

The Spark user community continued to expand. Databricks reported 30% growth in Spark Summit attendees and 240% growth in Spark Meetup members. 18% of respondents to Databricks’ annual user survey reported using Spark’s machine learning library in production, up from 13% in 2015. Among data scientists surveyed in the 2016 KDNuggets poll, 22% said they use Spark; in the 2016 O’Reilly Data Science Salary Survey, 21% of the respondents reported using Spark.

The Databricks survey also showed that 61% of users work with Spark in the public cloud, up from 51% in 2015. As of December 2016, there are Spark services available from each of the major public cloud providers (AWS, Microsoft, IBM and Google), plus value-added managed services for data scientists from Databricks, Qubole, Altiscale and Domino Data.

Apache Flink

dataArtisans’ Mike Winters reviewed Flink’s accomplishments in 2016 without using the words “machine learning.” That’s because Flink’s ML library is still pretty limited, no doubt because Flink’s streaming runtime is the primary user attraction.

While there are many use cases for scoring data streams with predictive models, there are few real-world use cases for training predictive models on data streams. Machine learning models are useful when they generalize to a population, which is only possible when the process that creates the data is in a steady state. If a process is in a steady state, it makes no difference whether you train on batched data or streaming data; the latest event falls into the same mathematical space as previous events. If recent events produce major changes to the model, the process is not in a steady state, so we can’t rely on the model to predict future events.

Flink does not yet support PMML model import, a relatively straightforward enhancement that would enable users to generate predictions on streaming data with models built elsewhere. Most streaming engines support this capability.

There may be use cases where Flink’s event-based streaming is superior to Spark’s micro-batching. For the most part, though, Flink strikes me as an elegant solution looking for a problem to solve.

Apache Mahout

The Mahout team released four double-dot releases. Key enhancements include the Samsara math environment and support for Flink as a back end. Most of the single machine and MapReduce algorithms are deprecated, so what’s left is a library of matrix operators for Spark, H2O, and Flink.

Apache OpenNLP

OpenNLP is a machine learning toolkit for processing natural language text. It’s not dead; it’s just resting.

Incubator Projects

In 2016, two machine learning projects entered the Apache Incubator, while no projects graduated, leaving six in process at the end of the year: SystemML, PredictionIO, MADLib, SINGA, Hivemall, and SAMOA. SystemML and Hivemall are the best bets to graduate in 2017.

Apache SystemML

SystemML is a library of machine learning algorithms that run on Spark and MapReduce, originally developed by IBM Research beginning in 2010. IBM donated the code to Apache in 2015; since then, IBM has committed resources to developing the project. All of the major contributors are IBM employees, which begs the question: what is the point of open-sourcing software if you don’t attract a community of contributors?

The team delivered three releases in 2016, adding algorithms and other features, including deep learning and GPU support. Given the support from IBM, it seems likely that the project will hit Release 1.0 this year and graduate to top-level status.

Usage remains light among people not employed by IBM. There is no “Powered By SystemML” page, which implies that nobody else uses it. IBM added SystemML to BigInsights this year, which expands the potential reach to IBM-loyal enterprises if there are any of those left. It’s possible that IBM uses the software in some of its other products.

Apache PredictionIO

PredictionIO is a machine learning server built on top of an open source stack, including Spark, HBase, Spray, and Elasticsearch. An eponymous startup began work on the project in 2013; Salesforce acquired the company earlier this year and donated the assets to Apache. Apache PredictionIO entered the Apache Incubator in May.

Apache PredictionIO includes many templates for “prebuilt” applications that use machine learning. These include an assortment of recommenders, lead scoring, churn prediction, electric load forecasting, sentiment analysis, and many others.

Since entering the Incubator, the team has delivered several minor releases. Development activity is light, however, which suggests that Salesforce isn’t doing much with this.

Apache SINGA

SINGA is a distributed deep learning project originally developed at the National University of Singapore and donated to Apache in 2015. The platform currently supports feed-forward models, convolutional neural networks, restricted Boltzmann machines, and recurrent neural networks.  It includes a stochastic gradient descent algorithm for model training.

The team has delivered three versions in 2016, culminating with Release 1.0.0 in September. The release number suggests that the team thinks the project will soon graduate to top-level status; they’d better catch up with paperwork, however, since they haven’t filed status reports with Apache in eighteen months.

Apache MADLib

MADLib is a library of machine learning functions that run in PostgreSQL, Greenplum Database and Apache HAWQ (incubating). Work began in 2010 as a collaboration between researchers at UC-Berkeley and data scientists at EMC Greenplum (now Pivotal Software). Pivotal donated the software assets to the Apache Software Foundation in 2015, and the project entered Apache incubator status.

In 2016, the team delivered three minor releases. The active contributor base is tiny, averaging three contributors per month.

According to a survey conducted by the team, most users have deployed the software on Greenplum database. Since Greenplum currently ranks 35th in the DB-Engines popularity ranking and is sinking fast, this project doesn’t have anywhere to go unless the team can port it to a broader set of platforms.

Apache Hivemall

Originally developed by Treasure Data and donated to the Apache Software Foundation, Hivemall is a scalable machine learning library implemented as a collection of Hive UDFs designed to run on Hive, Pig or Spark SQL with MapReduce, Tez or Spark. The team organized in September 2016 and plans an initial release in Q1 2017.

Given the relatively mature state of the code, large installed base for Hive, and high representation of Spark committers on the PMC, Hivemall is a good bet for top-level status in 2017.

Apache SAMOA

SAMOA entered the Apache Incubator two years ago and died. It’s a set of distributed streaming machine learning algorithms that run on top of S4, Storm, and Samza.

As noted above, under Flink, there isn’t much demand for streaming machine learning. S4 is moribund, Storm is old news and Samza is going nowhere; so, you can think of SAMOA as like an Estate Wagon built on an Edsel chassis. Unless the project team wants to port the code to Spark or Flink, this project is toast.

Machine Learning Projects

This category includes general-purpose machine learning platforms that support an assortment of algorithms for classification, regression, clustering and association. Based on reported usage and development activity, we cover H2O, XGBoost, and Weka in this category.

Three additional projects are worth noting, as they offer graphical user interfaces and appeal to business users. KNIME and RapidMiner provide open-source editions of their software together with commercially licensed versions; we cover these in Part Three of this survey. Orange is a project of the Bioinformatics Laboratory, Faculty of Computer and Information Science, University of Ljubljana, Slovenia.

Vowpal Wabbit gets an honorable mention. Known to Kaggleists as a fast and efficient learner, VW’s user base is currently too small to warrant full coverage. The project is now domiciled at Microsoft Research. It will be interesting to see if MSFT does anything with it.


H2O is an open source machine learning project of, a commercial venture. (We’ll cover’s business accomplishments in Part Three of this report.)

In 2016, the H2O team updated Sparkling Water for compatibility with Spark 2.0. Sparkling Water enables data scientists to combine Spark’s data ingestion and ETL capabilities with H2O machine learning algorithms. The team also delivered the first release of Steam, a component that supports model management and deployment at scale, and a preview of Deep Water for deep learning.

For 2017, plans to add an automated machine learning capability and deliver a production release of Deep Water, with support for TensorFlow, MXNet and Caffe back ends.

According to, H2O more than doubled its user base in 2016.


A project of the University of Washington’s Distributed Machine Learning Common (DMLC), XGBoost is an optimized distributed gradient boosting library used by top data scientists, who appreciate its scalability and accuracy. Tianqi Chen and Carlos Guestrin published a paper earlier this year describing the algorithm. Machine learning startups DataRobot and Dataiku added XGBoost to their platforms in 2016.


Weka is a collection of machine learning algorithms written in Java, developed at the University of Waikato in New Zealand and distributed under GPU license. Pentaho and RapidMiner include the software in their commercial products.

We include Weka in this review because it is still used by a significant minority of data scientists; 11% of those surveyed in the annual KDnuggets poll said they use the software. However, reported usage is declining rapidly, and development has virtually flatlined in the past few years, which suggests that this project may go the way of the eponymous flightless bird.

Deep Learning Frameworks

We include in this category software whose primary purpose is deep learning. Many general-purpose machine learning packages also support deep learning, but the packages listed here are purpose-built for the task.

Since they were introduced in late 2015, Google’s TensorFlow and Microsoft’s Cognitive Toolkit have rocketed from nothing to leadership in the category. With backing from Amazon and others, MXNet is coming on strong, while Theano and Keras have active communities in the Python world. Meanwhile, older and more mature frameworks, such as Caffe, DL4J, and Torch, are getting buried by the new kids on the block.

Money talks; commercial support matters. It’s a safe bet that projects backed by Google, Microsoft and Amazon will pull away from the pack in 2017.


TensorFlow is the leading deep learning framework, measured by reported usage or by development activity. Launched in 2015, Google’s deep learning platform went from zero to leadership in record time.

In April, Google released TensorFlow 0.8, with support for distributed processing. The development team shipped four additional releases during the year, with many additional enhancements, including:

  • Python 3.5 support
  • iOS support
  • Microsoft Windows support (selected functions)
  • CUDA 8 support
  • HDFS support
  • k-Means clustering
  • WALS matrix factorization
  • Iterative solvers for linear equations, linear least squares, eigenvalues and singular values

Also in April, DeepMind, Google’s AI research group, announced plans to switch from Torch to TensorFlow.

Google released its image captioning model in TensorFlow in September. The Google Brain team reported that this model correctly identified 94% of the images in the ImageNet 2012 benchmark.

In December, Constellation Research selected TensorFlow as 2016’s best innovation in enterprise software, citing its extensive use in projects throughout Google and strong developer community.

Microsoft Cognitive Toolkit

In 2016, Microsoft rebranded its deep learning framework as Microsoft Cognitive Toolkit (MCT) and released Version 2.0 to beta, with a new Python API and many other enhancements. In VentureBeat, Jordan Novet reports.

At the Neural Information Processing Systems (NIPS) Conference in early December, Cray announced that it successfully ran MCT on a Cray XC50 supercomputer with more than 1,000 NVIDIA Tesla P100 GPU accelerators.

Separately, Microsoft and NVIDIA announced a collaborative effort to support MCT on Tesla GPUs in Azure or on-premises, and on the NVIDIA DGX-1 supercomputer with Pascal GPUs.


Theano, a project of the Montreal Institute for Learning Algorithms at the University of Montreal, is a Python library for computationally intensive scientific investigation. It allows users to efficiently define, optimize and evaluate mathematical expressions with multi-dimensional arrays. (Reference here.) Like CNTK and TensorFlow, Theano represents neural networks as a symbolic graph.

The team released Theano 0.8 in March, with support for multiple GPUs. Two additional double-dot releases during the year added support for CuDNN v.5 and fixed bugs.


MXNet, a scalable deep learning library, is another project of the University of Washington’s Distributed Machine Learning Common (DMLC). It runs on CPUs, GPUs, clusters, desktops and mobile phones, and supports APIs for Python, R, Scala, Julia, Matlab, and Javascript.

The big news for MXNet in 2016 was its selection by Amazon Web Services. Craig Matsumoto reports; Serdar Yegulalp explains; Eric David dives deeper; Martin Heller reviews.


Keras is a high-level neural networks library that runs on TensorFlow or Theano. Originally authored by Google’s Francois Chollet, Keras had more than 200 active contributors in 2016.

In the Huffington Post, Chollet explains how Keras differs from other DL frameworks. Short version: Keras abstracts deep learning architecture from the computational back end, which made it easy to port from Theano to TensorFlow.


Updated, based on comments from Skymind CEO Chris Nicholson.

Deeplearning4j (DL4J) is a project of Skymind, a commercial venture. IT is an open-source, distributed deep-learning library written for Java and Scala. Integrated with Hadoop and Spark, DL4J runs on distributed GPUs and CPUs. Skymind benchmarks well against Caffe, TensorFlow, and Torch.

While Amazon, Google, and Microsoft promote deep learning on their cloud platforms, Skymind seeks to deliver deep learning on standard enterprise architecture, for organizations that want to train models on premises. I’m skeptical that’s a winning strategy, but it’s a credible strategy. Skymind landed a generous seed round in September, which should keep the lights on long enough to find out. Intel will like a deep learning framework that runs on Xeon boxes, so there’s a possible exit.

Skymind proposes to use Keras for a Python API, which will make the project more accessible to data scientists.


Caffe, a project of the Berkeley Vision and Learning Center (BVLC) is a deep learning framework released under an open source BSD license.  Stemming from BVLC’s work in vision and image recognition, Caffe’s core strength is its ability to model a Convolutional Neural Network (CNN). Caffe is written in C++.  Users interact with Caffe through a Python API or through a command line interface.  Deep learning models trained in Caffe can be compiled for operation on most devices, including Windows.

I don’t see any significant news for Caffe in 2016.