Hewlett Packard Enterprise last week announced the public availability of its HPE Haven OnDemand Machine Learning as a Service.
The Microsoft Azure cloud-based platform provides more than 60 APIs and services that deliver deep learning analytics on a variety of data, including text, audio, images, social Web and video.
Launched in beta in 2014, HPE Haven OnDemand has more than 12,750 registered developers generating millions of API calls per week, the company said.
It’s available as a free service for development and testing. Usage- and SLA-based pricing for enterprise-class delivery to support production deployment also are available.
“We’re bringing a unique solution to the market built on almost a decade of experience in advanced analytics and machine learning that has been proven,” said Jeff Veis, VP of marketing for big data at HPE.
“We have leveraged this experience into both the design and approach that we have adopted for Haven OnDemand,” he told the E-Commerce Times.
Haven OnDemand Features
Haven OnDemand offers Predict and Recommend capabilities that let developers view patterns in business data to optimize business performance and build a set of self-learning functions that analyze, predict and issue alerts based on structured data sets.
Advanced text analysis extracts the key meaning from language using concept extraction capabilities to obtain concepts and sentiment from text sources, HPE said.
Format conversion supports an extensive set of standard file formats and can employ optical character recognition to extract text from an image.
Other capabilities are Enterprise Search as a Service, image recognition and facial detection, knowledge graph analysis and speech recognition using advanced neural network technology to transcribe speech to text with support for more than 50 languages.
Developers worldwide been using Haven OnDemand through a global hackathon program, providing feedback and creating hundreds of applications, the company said.
Haven OnDemand is available globally through the Microsoft Azure public cloud, as part of a strategic alliance the companies announced last year.
Machine Learning Made Accessible
Haven OnDemand “has enabled us to democratize machine learning so any organization, large or small, can apply it,” HPE’s Veis said.
A lot of companies offer Machine Learning as a Service; Butler Analyticslists a dozen. Then there are Google’sTensorFlow, which has been open sourced; Microsoft’sAzure Machine Learning, available as a freemium service; and a slew of machine learning softwarelisted on Sourceforge.
“Leveraging machine learning was traditionally reserved for highly trained data scientists, academic institutions and premises-based software deployments,” Veis said. “Haven OnDemand is a breakthrough because we make enterprise-grade-level machine learning accessible to all developers with freemium pricing, self-service through the cloud.”
Market in Flux
HPE’s service “appears to compete with other, similar offerings in the market,” remarked Rob Enderle, principal analyst at the Enderle Group.
However, “this is a very new area, which means all these offerings are in flux and searching for that magical set of capabilities that folks want to use,” he told the E-Commerce Times.
The company that can create an offering with features the market wants and has the marketing budget to push the product will win, “because, right now, the big problem is the customers for these services don’t yet seem to know what they want,” Enderle said.
It’s not clear how well HPE Haven OnDemand will do because “HP isn’t really known to be a big player in this area, it’s new, and the initial showcase looks like it’s based on skills out of its printer group, which is no longer in HPE,” he said.
HPE “a company that should be able to figure it out and they know how to deliver solutions to enterprises, [but its] execution with new offerings has been pretty poor of late,” Enderle added.
“Until I can see them demonstrate an ability to move new products like this, I have to fall back on existing performance,” he said, “and that would suggest this will have limited success.”