Roundup 11/4/2016

Machine learning (ML) and deep learning (DL) content from the past 24 hours. Plus, some AI stuff.

The Next Platform publishes my three-part series on the state of enterprise machine learning. Part one is here. Part two is here. Part three is here.

HTAP for Machine Learning

— Monte Zweben, CEO of Splice Machine, summarizes the machine learning capabilities of Hybrid Transactional and Analytical Processing (HTAP) databases, including SAP HANA, Oracle Exadata, MemSQL, Splice Machine, Apache Hive, Apache HAWQ, Apache Trafodian, and Apache Impala with Apache Kudu. He thinks HTAP databases are great for ML.

For the record, here is a complete list of machine learning capabilities for Monte’s chosen platforms:

  • SAP HANA: Predictive Analytics Library, which has toylike ML tools
  • Oracle Exadata: nothing
  • MemSQL: nothing
  • Splice Machine: nothing
  • Apache Hive: nothing
  • Apache HAWQ: nothing native, but supports Apache MADlib
  • Apache Trafodian: nothing
  • Apache Impala with Apache Kudu: nothing

Other than that, it’s a good article.

Spark Summit Europe

Here are selected ML/DL highlights from the recent Spark Summit in Brussels. (Videos linked below; slides available here.)

— Ali Ghodsi: Democratizing AI with Spark

— Andy Steinbach: The potential of GPU-driven analytics in Spark

— Tim Hunter: TensorFrames: DL with TensorFlow on Spark

— Ali Zaidi: Scalable Bayesian inference.

— Nick Pentreath: Scaling factorization machines on Spark

— Rolf Jagerman: Glint, an asynchronous parameter server for Spark

— Dvir Volk: Accelerating Spark ML with Redis

— Debasish Das: Spark and Lucene for near-realtime predictive modeling

— Hollin Wilkins: Deploying ML to production

— Josef Habdank: Prediction as a service…

There are others, so check out the site.


— Matt Asay explains why AI and ML are hard. Answer: because the world is uncertain, and we strive to do hard things.

— Markets and Markets forecasts revenue from Machine Learning as a Service (MLaaS) to reach $3.8 billion by 2021. That may or may not square with IDC’s prediction that the broader market for cognitive systems and AI will expand to $47 billion in 2020. Separately, IDC projects spending on advanced and predictive analytics software to expand to about $4 billion.

— In Hackaday, Al Williams explains the foundations of machine learning.

— Google’s Greg Corrado explains machine learning basics.

— Ian Barker explains what you need to know about deep learning.

Methods and Techniques

— Brando Rohrer explains how Bayesian inference works.

— For $29, “Lazy Programmer” will teach you practical deep learning in Theano and TensorFlow. And $39 gets you the deep learning and artificial intelligence introductory bundle from “ScienceAlert Academy”.


— Adobe launches Sensei, a set of intelligent services running across Adobe creative, marketing, and document clouds. Linkapalooza here.


— In a surprising development, Kyoto University Graduate School of Medicine determines that a dual-socket Xeon system outperforms an NVIDIA K40 GPU when training deep neural networks for computational drug discovery using the Theano framework.


— A team of researchers uses machine learning to predict sex, height, weight, and foot size from hand dimensions, for some reason.

— In Forbes, an interview with an exec at Monsanto Growth Ventures, who discusses ML applications in agriculture.

— Professor Jackie Hunter discusses the use of DL in drug discovery.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.