Amazon Web Services, Microsoft team up on deep learning: Co-opetition is a driving force in the development of artificial intelligence technologies, evidenced by a new partnership between Amazon Web Services and Microsoft. The companies are collaborating on a new library for deep learning called Gluon. Here are the details from a Microsoft blog post:
Gluon is a concise, dynamic, high-level deep learning library, or interface, for building neural networks. It can be used with either Apache MXNet or Microsoft Cognitive Toolkit. Gluon offers an easy-to-use interface for developers, highly-scalable training, and efficient model evaluation–all without sacrificing flexibility for more experienced researchers. For companies, data scientists and developers Gluon offers simplicity without compromise through high-level APIs and pre-build/modular building blocks, and more accessible deep learning.
It's available on GitHub now, for use in conjunction with MXNet, with Cognitive Toolkit support coming later. AWS has also released a blog post with its take on Gluon, which is available here.
POV: The new partnership between Microsoft and AWS is reminiscent of the recent ONNX format announcement, in that it’s aimed at making it easier for developers to work with deep learning libraries, says Constellation VP and principal analyst Doug Henschen.
AWS embraced MXNet, a deep learning framework designed for cloud infrastructure, in May, and Gluon has already been a supported interface in MXNet.
At its recent Ignite event, Microsoft executives stressed that the company has an open strategy where deep learning and analytical frameworks are concerned, Henschen notes. "Efforts like ONNX and Gluon are aimed at making machine learning and deep learning work easier for a broader base of developers, but in both cases Microsoft’s favored library and those of its respective partners—Caffe2 in the case of the Facebook/ONNX partnership and MXNet in the case of the Amazon/Gluon partnership—is the first beneficiary."
Box adds AI to its content management mix: Content management unicorn Box used its Boxworks event this week to introduce Box Skills, a number of AI-powered capabilities designed to make the platform much more than a place to store and share files.
There are three initial skills. The first is Image Intelligence, which applies object detection along with handwriting and text recognition. The metadata created helps index images for search.
Audio Intelligence transcribes audio files, making them searchable by words or topics. Video Intelligence also provides transcriptions, plus topic detection and facial recognition.
Box has also created Skills Kit, which can be used by ISVs, SIs and enterprise IT shops to develop custom Box Skills, particularly for industry-specific business processes. The kit allows third parties to bring in whatever machine learning frameworks they'd like to use in conjunction with Box.
At BoxWorks, the company gave a few examples of custom skills, such as one giving quality assurance teams the means to search a call center recording database by topics and sentiment.
Finally, Box has unveiled Box Graph, a machine learning model that gains insight about a customer's organization according to how its workers use and share content. Initially, Box Graph will surface content an employee is or has been working on; recommend content from others it deems relevant; and show what content is the most popular within a company.
Both Box Skills and the Skills kit will enter beta early next year.
POV: "Bringing AI to the content in Box is important, because it elevates the content from just being stored in Box, to being used right within Box," says Constellation VP and principal analyst Alan Lepofsky.
It's notable that Box has architected the Skills kit to enable a variety of machine learning models, he adds. However, it's not yet clear what the licensing and pricing will be for associated API calls, Lepofsky notes.
Check out Lepofsky's more in-depth take on Box's news in the video below.
Examining Dell's billion-dollar IoT bet: Dell Technologies has announced it will spend $1 billion on product development, partnership activities and other areas over the next three years as part of a new division centered on IoT. It is the company's biggest push yet into IoT. Here are some key details from Dell's announcement:
The company’s new IoT Division will be led by VMware CTO Ray O’Farrell, and is chartered with orchestrating the development of IoT products and services across the Dell Technologies family. The IoT Solutions Division will combine internally developed technologies with offerings from the vast Dell Technologies ecosystem to deliver complete solutions for the customer.
In essence, the division will bring together software and hardware assets Dell already has, such as VMWare Pulse IoT Control Center and Edge Gateways, while bringing in OT (operation technology)-centric products from partners.
Dell is also working on a number of new products, such as Project Nautilus, a software platform that ingests and analyzes IoT gateway data in real time, and Project IRIS, for security at the edge of IoT networks.
POV: Dell's desire to be a one-stop shop for IoT, albeit with the help of partners, reflects the current market landscape. Hitachi recently launched Vantara, an IoT-focused unit that it says provides both IT and OT know-how.
It's also important to look at the bigger picture underlying Dell and others' IoT moves.
"There is a new accord developing across the market around the manner in which IoT and AI—and directly associated technologies such as machine learning—are deployed to deliver business value," says Constellation VP and principal analyst Andy Mulholland. "Increasingly, the term systems of Eegagement is used to describe the manner in which data is gathered around the enterprises activities in respect of the real world of events and actions, whereas the term systems of record is used to describe the role of Enterprise IT in recording transactions to create historical data for analysis."
The point is that just as enterprise IT aligns to the business objective of centralizing and reducing variation in process, systems of engagement align to the new digital business model of engaging with the events at the ‘edge’ of the enterprise, where it interacts with its markets, customers, suppliers, even its own dynamic operations, he adds.
"As deployment experience builds it has become very clear that a significant amount of edge activity must be processed at the edge, within the context and timeframe that drives its business value," Mulholland says. "Cisco identified this as fog computing several years back, but Dell has benefited from its significant market in industrial processors connected to sensors on automated production lines to really grasp the requirement and with this investment move to take a leading position."
