Google's Artificial Intelligence Successfully "Identifies" Cat
News arrives from Google's esoteric X Lab that its Artificial Brain has successfully learnt to identify a cat. The Artificial Intelligence project which is a neural network comprising of 16,000 processing cores, simulates the human brain.
The Artificial Brain was sent off on a task which involved browsing through innumerous cat videos online, a task most of us associate with, among others. The system was supplied with 10 million thumbnail images taken from YouTube as “training†data by a team from Google X and Stanford University. Of the available database, the system was tested if it could recognize 20,000 objects in new images.
#-Link-Snipped-#
The Artificial Intelligence system performed with an accuracy twice as of any previous neural network. The viral popularity of cat videos made for an abundant database for the system to learn from. Without any prior information provided to identify a cat, the system was able to invent its own concept of cat derived from the research.
Though Google's neural network is yet quite far from operating on the level of a human visual cortex, this research proves that the Artificial Brain can learn with little "help" provided it has a large dataset at hand.
Source: #-Link-Snipped-#Â Image Credit: #-Link-Snipped-#
The Artificial Brain was sent off on a task which involved browsing through innumerous cat videos online, a task most of us associate with, among others. The system was supplied with 10 million thumbnail images taken from YouTube as “training†data by a team from Google X and Stanford University. Of the available database, the system was tested if it could recognize 20,000 objects in new images.
#-Link-Snipped-#
The Artificial Intelligence system performed with an accuracy twice as of any previous neural network. The viral popularity of cat videos made for an abundant database for the system to learn from. Without any prior information provided to identify a cat, the system was able to invent its own concept of cat derived from the research.
Though Google's neural network is yet quite far from operating on the level of a human visual cortex, this research proves that the Artificial Brain can learn with little "help" provided it has a large dataset at hand.
Source: #-Link-Snipped-#Â Image Credit: #-Link-Snipped-#
Replies
You are reading an archived discussion.
Related Posts
In a combined effort by McAfee and Guardian Analytics, the security firms on Tuesday released the report on Operation High Roller which has been delving into the details of fraudulent...
Relativistic Heavy Ion Collider (RHIC), an atom-smasher, just made it to the Guinness book of records for having produced “the hottest man-made temperature ever”. How hot could it be? Real...
According to the reports, Amazon is readying its announcement for the Kindle Fire 2 in a press event on July 31st. The Kindle Fire 2 is believed to bear the...
Tony Hoare is a principal researcher from Microsoft Research in Cambridge. Fascinated by the power of mathematical logic, he has reworked the famous 'Turing Test' for programmers. Speaking at Alan...
Ishikawa Lab, Japan has developed a robot called janken that plays rock-paper-scissors with you and no matter how worse you feel about it, it beats you every single time. The...