Machine Learning Roundup: 10/5/2016
Machine learning (ML) and deep learning (DL) content from the past 24 hours, plus a few older items. Top stories today: open standards for neural networks; continuing coverage of Australia’s investment in deep learning supercomputers; individual sections on Health Tech and Factory Automation, plus some top-rated books (other than mine.)
As always, Adrian Colyer provides the top read of the day: Why does deep and cheap learning work so well?
From Last Week
In Quanta Magazine, Ingrid Daubechies explains how art conservators use mathematical analysis to restore a masterpiece. Not exactly machine learning, but a good read. (h/t Oliver Vagner)
Open Standards for Neural Networks
The Khronos Group, a consortium of hardware and software companies, announces two initiatives to promote the development of neural network techniques. The Neural Network Exchange Format (NNEF) initiative will develop an open standard file format to exchange deep learning models between training and inference systems. The OpenVX Neural Network Extension project will be a high-level architecture specification to run Convolutional Neural Networks (CNN) as OpenVX graphs. Brandon Lewis reports.
Ozzies Deploy Deep Learning Supercomputers
Australia’s Commonwealth Scientific and Industrial Research Organization (CSIRO) deploys two stupidly powerful NVIDIA DGX-1s, each of which has the throughput of 250 servers. First up on the list of projects: sifting through massive quantities of data to understand the impact of environment on disease. In ZDNet, Chris Duckett reports.
ML/DL in Health Tech
In a press release, Orion Health touts its use of machine learning to sift through data.
Startup Healx, based in Cambridge (U.K.), lands a Series A round of a little less than $2 million. The company uses ML to find new uses for existing drugs.
In Health Data Management, Arthur Layne opines that AI and ML can reduce data breaches.
Dallas-based DocSynk closes on a $1 million seed round. DocSynk uses ML to match patients and doctors.
Chicago-based Tempus announces its selection by the Pancreatic Cancer Action Network. Tempus will collect and analyze genomic data for a field trial of Precision Promise, a large-scale initiative to improve health outcomes for pancreatic cancer patients.
Researchers at the University of Edinburgh use machine learning to predict strains of bacteria likely to cause food poisoning outbreaks. Joe Whitworth reports.
ML/DL in Factory and Workplace Automation
In Tokyo, NVIDIA and FANUC announce an alliance to embed DL-driven AI in the FANUC Intelligent Edge Link and Drive (FIELD) system for robotics. Adding AI enables robots to teach themselves to perform tasks more efficiently. FANUC will use NVIDIA GPUs and DL software.
In VentureBeat, TopBots’ Adelyn Zhou describes how AI and ML disrupt the workplace.
Sheila Kennedy surveys ML-driven innovation in manufacturing, including process optimization, maintenance optimization, and failure analysis.
Here are three pertinent books with 5-star customer reviews on Amazon:
- Kelleher, MacNamee, and D’Arcy: Fundamentals of Machine Learning for Predictive Data Analytics
- Chou, Precision: Principles, Practices, and Solutions for the Internet of Things
- James, Witten, Hastie and Tibshirani: An Introduction to Statistical Learning
John Paul Mueller’s Machine Learning for Dummies gets only 3 1/2 stars, but it’s mandatory for IBM executives.
In Nature, Davide Castelvecchi asks: can we open the black box of AI?
Hannah Kuchler, writing in Financial Times, describes the AI/ML arms race between security providers and hackers.
DataKind founder Jake Porway thinks we should use data for his pet causes and not “just sell stuff.” Looking at DataKind’s numbers, it looks like DataKind could use a little more selling. If only they had stuff to sell, other than their founder’s opinions.
Computerworld’s Martin Heller reviews TensorFlow and likes it.
Salesforce CMO groks machine learning.
In Geekwire, Taylor Soper surveys Microsoft Ventures’ startup investments this year. Two focus on machine learning:
- CognitiveScale offers a “Cognitive Cloud” built on an open platform, with embedded deep cognition, computer vision, and machine learning.
- CrowdFlower combines a collaboration platform and data science marketplace.
Episerver, a “digital experience provider,” acquires Berlin-based optivo, which provides marketing services based on multidimensional segmentation and event triggering. Episerver plans to integrate optivo’s technology with a recently acquired predictive analytics platform and decision engine.
Phil Wainewright asks if Salesforce Einstein is rocket science, concludes that it is.
Joel Shore argues the developers should become experts in AI. If that happens, they won’t be developers anymore.
Den Howlett interviews Infosys CEO Vishal Sikka, who seeks to demystify AI, ML, and DL.