HPE Launches On-Demand Machine Learning-As-A-Service

‘We refer to this as ‘augmented human intelligence,’ so they can generate applications and build these features into their products.’ – HPE’s Colin Mahony

Hewlett Packard Enterprise has joined a growing number of companies—such as AWS, Microsoft Azure, startups Nervana and BigML, Splunk and Salesforce—who see dollar-sign potential in making machine-learning analytics easily available to software developers as a cloud service.

HPE on March 10 launched Haven OnDemand, which provides advanced machine-learning application programming interfaces (APIs) and services that enable developers, startups and enterprises to build next-gen mobile and enterprise applications.

No more are artificial intelligence and algorithmic analytics the exclusive purview of The Data Scientist Club. The fact is, there are not nearly enough data scientists in the world to determine the meaning of all the data that companies are now storing, so this type of application is basically an answer to a cry for help from the IT world.

Available on Azure Cloud

microsoftHPE’s Haven OnDemand, at this early point, is delivered as a service exclusively on Microsoft Azure. It contains more than 60 APIs and other microservices—with many more expected to come—that enable deep-learning analytics on a range of data; these include text, audio, image, social, web and video.

“We refer to this as ‘augmented human intelligence,’ so developers can generate applications and build these features into their products,” HPE Vice President of Big Data Colin Mahony told eWEEK.

HPE also launched a search engine, Haven Search OnDemand, that goes hand-in-hand with the Haven OnDemand service.

HPE (then it was just plain HP) launched the beta version in December 2014. Since then, the platform has attracted 12,750 registered developers to date who currently generate millions of API calls per week and have provided valuable feedback to improve and refine the platform, Mahony said.

Haven OnDemand is a freemium service that enables testing and development for free, and then expands as needed to a usage and SLA-based commercial pricing model for enterprise class delivery to support production deployments.

Some of the capabilities and APIs offered by HPE Haven OnDemand include:

Advanced Text Analysis: Extracts the key meaning from language by employing powerful concept extraction capabilities that go beyond traditional approaches to obtain key concepts, entities and sentiment from text sources.

Format Conversion: Provides key functions to access, extract and convert information wherever it lives by supporting an extensive set of standard file formats and the ability to employ optical character recognition to extract text from an image.

Search OnDemand service: Delivers powerful cultivated search across on-premise or cloud data to deliver superior context-sensitive search results.

Image Recognition and Face Detection: Enables applications to detect specific image features and code around human-centric use cases to identify the gender of an individual or key information such as a brand logo from within an image.

Knowledge Graph Analysis: Automatically delivers insights and predictions related to relationships and behavioral patterns among people, places and things. These capabilities are very useful for analyzing social media and related data.

Predict and Recommend: Enables developers to view patterns in business data to optimize business performance and build a set of self-learning functions that analyze, predict and alert based on structured datasets.

Speech Recognition: Employs advanced neural network technology to transcribe speech to text from video or audio files with support for over 50 languages.

Use Case: Social Capital

eWEEK spoke with Tope Alibi, co-founder of San Francisco-based startup Social Capital, whose analytics-based service helps HR departments find good job candidates by trolling social networks and identifying appropriate personality traits for the jobs. Alibi was a beta program user of Haven OnDemand.

“Our algorithm can handle any type of digital media, which could be email, Slack messages, Skype messages, whatever, but we started with social because it’s free and publicly available,” Alibi said. “Haven OnDemand has been a great tool for us. With it, we identify the ‘big five’ personality traits that psychologists identify with each patient: openness, conscientiousness, extraversion, agreeableness and neuroticism.

“We do this by finding reflective writing; we find which adjectives you use in describing yourself, your dog, or your friends. We search for functional words that describe how you are communicating, how you are thinking through these things. It’s not what they are saying, it’s how they are saying it.

“We’re in love with the problem of understanding people. This is a general problem. Whether you’re talking about dating, hiring, or just you and your brother, people want to understand other people. OnDemand runs this whole thing for us.”

Originally published on eWeek.