GOOGLE CREATES ITS OWN ‘AI CHILD’
A Google super computer is able to over perform humans and now it’s time for the ‘AI Child’. NASNet is able to identify objects like people, videos, images and according to reports Google “brain” can select these objects with 82.7% accuracy rates which is predicted to be better than man-made AI system.
So break-through of AI is certain to have a huge impact on the entire field. Automation is creating more effective and accurate AI systems with meaningful intelligence incorporated in them.
NASNET has already created AutoML this year which is an AI system capable to learn more programmes. The Google brain team, who has been involved in creating AutoML, hopes that the larger machine learning community will be able to execute this model to address the computer vision. Technology has always been on a double edged sword but with AI it is going to be more empowering to human simplifying things in a much better way.
A controller neural can propose a child model that could be trained and evaluated on a quality check to a particular task. NASNet has incorporated AI for object detection to outperform the state of art of machine learning architectures build in favour of academic competitions by human.
Google has applied to ImageNet classification and COCO object detection which describes two of the large scale academic data sets in computer vision. With large machine learning community there will be ability to build on these models to address multitude of computer visions which was never imagined earlier.