Home > Geek

Google Brain Algorithm has brought digital picture enhancement to the next level

By Menahem Zen | Feb 10, 2017 03:58 AM EST
The new Google logo is displayed on a sign outside of the Google headquarters on September 2, 2015 in Mountain View, California.
(Photo : Getty Images/Justin Sullivan) The new Google logo is displayed on a sign outside of the Google headquarters on September 2, 2015 in Mountain View, California. Google has made the most dramatic change to their logo since 1999 and have replaced their signature serif font with a new typeface called Product Sans.

The new technology from Google will enable people to do things which can only be seen in crime TV series. Google Brain algorithm can enhance digital picture and superzoom the picture.

Such technology which can only be imagined in the TV series such as “CSI” or crime investigation film will soon be available. According to Ars Technica, the new algorithm from Google Brain will be able to do exactly like that. The new software can enhance a picture from a tiny and pixelated source images into detailed images.

The new algorithm make a combination of two neural networks to work together, they are conditioning network and prior network. The conditioning will map the source image and match the image with high resolution images available. Afterward, the prior network will deploy the PixelCNN image decoders to create a high resolution details of the image.

Google called the latest development is able to create what they called a “pixel recursive super resolution," Gizmodo reported. Combination between the two algorithms is able to conduct a super zoom-in and enhance the image. Nevertheless, the technology still need an enhancement until able to make a real prediction.

Google Brain is the division in Google which established as independent project in 2011 as the deep research division. The main focus of the division is to conduct an Artificial Intelligence and machine learning research.

At first, Google Brain was established as research collaboration between three top professors from Stanford University and Google Research team. They are Google Fellow Jeff Dean, Google Researcher Greg Corrado, and Stanford University professor Andrew Ng. Other researchers join later on including Greg Corrado, Quoc Le, Alex Krizhevsky.

Two prominent deep learning professors Ilya Sutskever and Geoffrey Hinton also contributed to the project. Sutskever is an Open AI Research Director while Hinton is a prominent figure in the Artificial Intelligence and deep learning with his theory of generalized backpropagation algorithm to create a multi-layer neural networks.

Watch the explanation from Senior Fellow in the Systems and Infrastructure Group at Google Jeff Dean regarding the Google Brain and Brain Residency from Google Student YouTube channel below:

© Copyright 2016 AsiaStarz.com. All Rights Reserved.


Real Time Analytics