Google launches service to make machine learning easier

20160323 jeff dean gcpnext

Jeff Dean discusses Google TensorFlow at the company's GCP Next conference in San Francisco on March 23, 2016.

Credit: Blair Hanley Frank

The Cloud Machine Learning service makes it faster to create intelligent applications with Google's cloud

Google is making it easier for businesses to take advantage of the machine learning revolution with a new product for building models that predict the future. 

At the company's GCP Next conference in San Francisco on Wednesday, Google announced the private beta of a new Cloud Machine Learning service that lets businesses create a custom machine learning model. To do so, users work with data they have in Google's other cloud services. Cloud Machine Learning handles data ingestion and training and then uses the resulting machine-learning model to make predictions. 

It's designed for companies that want to use machine learning to make predictions for their business. Jeff Dean, the head of Google's Brain deep-learning research project, showed the service making a model that predicted when a user would click on an advertisement. It was based on anonymized data from marketing software company Criteo about when users click. 

Cloud Machine Learning is built on top of Google's open-source TensorFlow product, which also provides the intelligent capabilities inside many of Google's other products, such as Inbox and Photos. It's the most popular machine learning project on GitHub and helped underpin the AlphaGo A.I. that recently beat a human champion at the board game Go. 

The tools are key to Google's staying competitive in the cloud market. Both Microsoft Azure and Amazon Web Services already have managed machine learning services designed to make it easier for people to build machine learning models and then consume them in the cloud. 

One of the things that eventually will set Cloud Machine Learning apart is that Google will let companies export their TensorFlow models from its cloud and use them in other settings, including on-premises datacenters.

Right now, both Azure's and AWS's machine learning products require people to consume the machine-learning models they've built in the cloud through an API, which locks users of those products into each company's ecosystem. 

Also on Wednesday, the company announced a new Cloud Speech API that allows users to take in snippets of speech and have them transcribed based on the machine learning model that underpins Google's speech recognition technology.

It's part of a growing set of machine learning APIs that Google has released, which also includes its Cloud Vision API (which will return insights about the content of an image) and Cloud Translate API.

More broadly, there are plenty of other companies, including Microsoft and IBM, trying to capitalize on a growing API economy that sells companies intelligent computing capabilities to use in their applications. 

Google says these capabilities will lure companies interested in building smart applications to its cloud and not others, especially because of the company's track record of machine learning innovation. What remains to be seen is whether that works, especially since Google is lagging behind Microsoft and Amazon in public cloud use. 

Lock down your servers more easily: A look inside the Microsoft Local Administrator Password Solution
View Comments
Join the discussion
Be the first to comment on this article. Our Commenting Policies