Google open-sources machine learning to smarten up our apps

09.11.2015
One day in the not-so-distant future, an app might make a dinner reservation for you before you realize you even want to go out, or your smartphone might suggest tourist sights you'd enjoy when you land in a new city.

It's possible -- and it's really not so far away, say analysts, who were encouraged today by Google's announcement that it's open sourcing an enhanced machine learning system.

The system, dubbed TensorFlow, is smarter, faster and more flexible machine-learning software than Google has ever had before, according to Sundar Pichai, Google's CEO, in a blog post .

"Just a couple of years ago, you couldn't talk to the Google app through the noise of a city sidewalk, or read a sign in Russian using Google Translate, or instantly find pictures of your Labradoodle in Google Photos," wrote Pichai. "Our apps just weren't smart enough. But in a short amount of time they've gotten much, much smarter. Now, thanks to machine learning, you can do all those things pretty easily, and a lot more."

Now the new machine learning system should take intelligent systems even further.

TensorFlow has been designed to adapt more easily to new research and new applications, according to Google.

"It's a highly scalable machine learning system," Pichai noted. "It can run on a single smartphone or across thousands of computers in data centers. We use TensorFlow for everything from speech recognition in the Google app, to 'smart reply' in Inbox, to search in Google Photos. It allows us to build and train neural nets up to five times faster than our first-generation system, so we can use it to improve our products much more quickly."

And now other artificial intelligence researchers and engineers can use TensorFlow because Google is open sourcing the system.

"Open sourcing it gives developers of all sizes access to machine learning, making their apps smarter," said Zeus Kerravala, an analyst with ZK Research. "Open sourcing it could be very helpful because the community together will think up use cases that Google or any single developer could not."

Ezra Gottheil, an analyst with Technology Business Research, noted that better machine learning could help companies figure out when and where their products are likely to break down.

It's all about making better use of big data and enabling apps, robots and networks learn and grow on their own.

Machine learning is a field inside the larger arena of artificial intelligence (AI).

Artificial intelligence is about helping machines perform tasks that traditionally required human intelligence, including decision-making and speech translation. Machine learning, however, is focused on building algorithms that can learn from and make predictions based on information being fed into them.

Many companies, like IBM with its Watson system, and universities including Carnegie Mellon University and Stanford University are working on machine learning and AI.

(www.computerworld.com)

Sharon Gaudin

Zur Startseite