The Emergence of Machine Learning
Microsoft is hoping to boost the accessibility of machine learning technology with a new predictive analytics cloud service, Microsoft Azure Machine Learning. The service is part of a movement that includes a new breed of business intelligence vendor that aims to help businesses take advantage of the increasing volume of data they collect.
The new service is one of several machine learning technologies that have emerged in the past few years. It shows the need to manage data and the predictive analytics technologies that are increasingly required to analyze complex data structures.
Microsoft said it built the new service using technologies that it developed for other products like Xbox and Bing. The service will include “visual workflows and startup templates” designed to make it easy for users to set up machine learning tasks, Joseph Sirosh, corporate vice president of machine learning at Microsoft, wrote in a blog post about the development.
He’s right when he says that setting up machine learning processes is tough. That’s especially true today when data scientists are in high demand.
Some businesses are trying to use BI products to combine data they collect from many different sources and do predictive analytics in order to better anticipate product repairs, for instance, or customer behavior. There are many new BI products on the market that are designed to make data analytics much easier to use so that more people within a business can use them. Companies like Birst, Good Data and Tableau are some of this new breed of BI vendor.
But few of the BI products do predictive analytics or if they do, not elegantly. In most cases, engineers have to do a lot of custom work to develop projects that do predictive analytics. That means they aren’t able to be very agile, setting up new scenarios quickly to learn new things.
Microsoft hasn’t revealed a lot about its service, which it said is being used by some customers in an early preview and which it said it plans to release as a public preview in July. Sirosh told the New York Times that the service will allow developers to use R, a popular open source programming language for doing machine learning. He also told the paper that around 100 companies and universities were already using the service.
Mary Jo Foley at ZDNet got ahold of a few screen shots that show what the service looks like.
If the new Azure Machine Learning service manages to do what Microsoft hopes, businesses are sure to welcome it.
That said, Microsoft isn’t alone in trying to move beyond the custom BI-based implementations. Earlier this year, IBM started making available a service that lets users access its Watson supercomputer machines in the cloud in a way that was designed to make it easier for businesses to run machine learning applications.
Amazon Web Services’ Kinesis service can also be used for machine learning. Kinesis is designed to allow for the processing of real-time streaming data.
The New Stack Founder Alex Williams, when writing for TechCrunch, described Kinesis as a service, designed for real-time apps, allows developers to pull any amount of data, from any number of sources, scaling up and down as needed.
He wrote that “Kinesis can create any number of streams across multiple availability zones. The streams have no “intrinsic” capacity or rate limits. All incoming data is replicated across availability zones for high availability.
It’s a safe bet that Google is probably also working on a similar kind of offering, given the demand for tools that can help businesses make sense of their data. It currently offers a predictive analytics API.
Plus, there are the number of open source projects that have emerged including Apache Mahout and Predictionio, a prediction server for building smart applications, which has more than 4.600 stars on GitHub.
Feature image via Creative Commons.