HPE takes AI to the cloud with machine learning-as-a-service

HPE recently made its machine learning cloud platform, Haven OnDemand, generally available. The service offers 60+ APIs for developers to add analytics to their apps.

Image: iStockphoto/tonefotografia

On Thursday, HPE announced the immediate commercial availability of HPE Haven OnDemand, a cloud platform that provides advanced machine-learning APIs, so that developers can build data-rich mobile and web applications.

Haven OnDemand was initially launched in December 2014, but it has been in limited preview until now. The APIs available through Haven OnDemand cover the bases of deep learning, text and audio recognition, image analysis, social sentiment, video analysis, facial detection, speech analysis, and more. The platform primarily offers RESTful APIs--APIs that adhere to the six architectural constraints of representational state transfer (REST)--and there are currently more than 60 APIs and services available.

Colin Mahony heads up the HPE Big Data Platform, which includes products like Vertica, Idol, and Haven OnDemand. He said that some people in the enterprise don't embrace machine learning because they view it as requiring extensive understanding of both statistics and coding. And, with Haven OnDemand, they want to make it easier for people to take advantage of machine learning tools.

SEE: Job description: Data scientist (Tech Pro Research)

"What we're trying to do is say here's a portfolio, initially these 60 APIs, where you can call in, in very simple protocols through these RESTful APIS, and you can leverage a lot of the really rich machine learning that we've done," Mahony said.

The service, which currently has 12,750 registered developers, is delivered on Microsoft Azure and, while it has been around for a while, will bring a few new features about with its commercial availability. Some of the key features of the platform include:

  • Advanced text analysis
  • File format Conversion
  • Haven Search OnDemand enterprise search as a service
  • Image recognition and face detection
  • Knowledge graph analysis
  • Predict data patterns and recommend options
  • Speech recognition

Making Haven OnDemand generally available is the latest in a string of analytics updates that HPE has been making over the past year. Back in February the company launched its Investigative Analytics tool for financial organizations, and a few weeks later it upgraded its AppPulse analytics suite as well.

The group of announcements points to fact that data analytics and AI tools will play a big role in the future of HPE and its product offerings. Mahony went as far as to say, to him, "analytics is everything."

"Having seen what so many companies have been able to do with our analytics platforms, and frankly with others, it is incredible to see what a difference analytics makes on the decisions that organizations can do," Mahony said.

SEE: HPE launches Investigative Analytics, using AI and big data to identify risk (TechRepublic)

In addition to making tools like Vertica, Idol, and Haven OnDemand available in their own right, Mahony said that HPE sees its analytics capabilities playing a role in their other products as well, and they are tools that are being used internally at HPE too.

While HPE would love to be a one-stop shop for analytics, and it is positioning itself as just that, Mahony said that the company is committed to open standards, an open ecosystem, and open source. Not all of its products are open source though, but HPE is keeping the ecosystem open to allow customers to choose a different partner option if they so desire.

Haven OnDemand will start as a freemium model, with upper pricing tiers dependent on the number of API calls and resource units. The basic Explorer model includes one resource unit and 20,000 API units for $10 per month. The Innovator level offers 10 resource units and 50,000 API units for $85 per month, and the top tier Entrepreneur model gives users 106 resource units and 50,000 API units for $315 a month. While these are the standard pricing models, Mahony said that HPE will sometimes work with organizations to come develop a custom plan.

Interested parties can find more about Haven OnDemand here.

The 3 big takeaways for TechRepublic readers

  1. HPE made Haven OnDemand, its cloud platform for machine learning, generally available. Developers can use the 60+ APIs to add data analytics features to their applications.
  2. Haven OnDemand requires little existing knowledge of the intricate working of data analytics and machine learning to be useful. Companies without a dedicated data scientist could use Haven OnDemand to begin experimenting with machine learning and analytics.
  3. HPE has made a number of analytics announcements in the past few months, and it seems to be focusing heavily on those capabilities throughout its product line. Expect to see more analytics and AI tools from HPE in the future.

Also see