BT

Facilitating the Spread of Knowledge and Innovation in Professional Software Development

Write for InfoQ

Topics

Choose your language

InfoQ Homepage News Microsoft Launching Azure Machine Learning as a Service

Microsoft Launching Azure Machine Learning as a Service

Bookmarks

Microsoft recently announced Azure ML, a machine learning cloud based platform that helps predict future events based on past performance. Microsoft has been using machine learning for years for Bing, Xbox and other products but this is the first time that internal technologies are consumerized and deployed as cloud services.

Azure ML is bringing an intuitive user friendly web interface that understands the R language and already features more than 350 R packages ready to be used by data scientists. Azure ML can utilize data from a wealth of inputs including Microsoft’s Hadoop as a Service HDInsight , SQL data sources, Azure storage and even Azure Virtual Machines.

Oriented for the data scientists rather than the software engineers, Azure ML users can drag and drop visual components to create the data flow and experiment that they need to run. ML studio also has predefined experiments and algorithms that are battle tested by Microsoft in Xbox, Bing and Research departments.

Early Azure ML customers feature Max451 and Carnegie Mellon University. OSISoft is currently working with Carnegie Mellon University to analyze fault detection in real time and forecast energy needs across campus buildings.

Machine Learning is a hot field with all major technology companies heavily investing in it, as previously featured by InfoQ. The difference though is that instead of providing pure API’s to be used by developers like Google’s prediction API does, Azure ML and other efforts are moving to the state where data scientists can ingest data, analyze them and prototype API’s themselves.

Ersatz Labs, a San Francisco based startup is also trying to address Machine Learning from the Data Scientist’s perspective. Their ML web based platform deals with data wrangling and offers predefined machine learning algorithms to be used by data scientists towards their goal. Using GPU powered hardware they claim to achieve faster execution and their focus is to enable users to use their API relatively easily to develop applications.

While still in preview, Azure ML doesn’t offer an SLA and pricing is including a preview discount. Machine Learning as a Service is on its way to be democratized and this is going to help data analysts to develop services easier with a shorter time to market.

Rate this Article

Adoption
Style

BT