Last week, HP officially launched commercial plans for Haven OnDemand, a machine learning platform that offers developers a suite of APIs they can use to "turn virtually any data into an asset anytime and anywhere." The platform entered beta in 2014 and offers more than offers more than 60 APIs.
The platform offers more than 60 APIs for a variety of machine learning-related applications. These include APIs for speech recognition, anomaly detection, image and facial recognition, text analysis, search and predictive modeling. For example, using the predictive modeling API, developers can create prediction models according to training data sets. The API includes a number of algorithms, such as decision tree, logistic regression and naive Bayes, and applies the algorithm it determines is best.
Many of Haven OnDemand's JSON REST APIs can be used synchronously and asynchronously, giving developers the flexibility to incorporate Haven OnDemand's functionality into their applications using both real-time and non-real-time approaches. The APIs are offered under a subscription model and based on usage as measured by Resource Units and API Units. The former correspond to index sizes and models, while the latter correspond to API calls.
HP offers a free tier off 10,000 API Units, while its Explorer, Innovator and Entrepreneur packages range in price from $10 to $315 per month for up to 120,000 API Units. To celebrate its commercial launch, HP is running an introductory promotion that increases the Resource Units and API Units for each paid plan in the first three months.
Fernando Lucini, the VP of HP Haven OnDemand, says that HP "believe[s] that the possibilities of what developers can do with Machine Learning are endless and that we’re only at the beginning the journey." With this in mind, the company made the decision to launch Haven OnDemand under a freemium pricing model to ensure that it is accessible to "developers at all stages of their growth, from early stage startups to global enterprises."
According to Lucini, more than 12,000 developers have joined the Haven OnDemand developer community since 2014 and the platform's services are already being used by a number of companies. These include Ayni, a startup that uses Haven OnDemand's speech recognition API and RingDNA, which uses Haven OnDemand's APIs in the conversation analytics portion of its sales solution.
Haven OnDemand also powers Haven Search OnDemand, a commercial enterprise search-as-a-service solution that HP offers.
Machine Learning on the Rise
According to Geoffrey Gordon of Carnegie Mellon University’s Machine Learning Department, "At its highest level, machine learning is about understanding the world through data."
With data volumes growing by leaps and bounds, interest in machine learning is skyrocketing, as evidenced by soaring enrollment in machine learning classes at universities. As Nvidia's Jamie Beckett detailed, the undergraduate Introduction to Machine Learning course at Carnegie Mellon University has increased by almost 600% in the past five years.
Given the growing need to make sense of large amounts of data, it's no surprise that there are a growing number of platforms offering machine learning as a service and that, increasingly, some of the largest players in tech are active in the space. In addition to HP, Google has a Prediction API, Amazon offers its own Machine Learning service as part of AWS, and last year, IBM added AlchemyAPI, which offers a collection of machine learning APIs, to the IBM Watson portfolio.
All of the activity and investment in this space is good news for developers, who should benefit as companies seek to innovate and ensure their offerings are competitively priced.