A team of researchers with one of Indian-origin have developed a postage-stamp-size chip which would be able to simulate numerous the human brains neurons. Such 'optical neural networks' may create any application of so-called deep learning. These are from virtual assistants to language translators who are much quicker and more efficient.

HIGHLIGHTS
Researchers have developed a new chip to stimulate neurons of the human brain
New chip made of silicon simulate a network of 16 neurons in four 'layers' 
New chip is more efficient than computer chip and it produces less heat

Also Read: Musk - Human Brain Soon To Gain Artificial Intelligence!

The chip which stimulated the human brains were eventually equipped with 5.4 billion transistors and is capable of simulating one million neurons and 256 million neural connections, or synapses. Along with mimicking the human brain's functioning by themselves, individual chips could get often connected along like tiles, almost like the way circuits are joined within the human brain.

It was found that the new chip was made of silicon material which would simulate a network of around 16 neurons in four 'layers' of four. It has been seen that the data enters the chip like a laser beam which would split into four smaller beams. The brightness of every beam which is entering into beam signifies a special number, or piece of information, and therefore the brightness of every exiting beam would represent a brand new number, the 'solution' after the data has been processed. Well, in between that the paths of light crosses and then interact in ways in which it can amplify or weaken their individual intensities, in the constant way ocean waves could add or substract from one another after they cross. These crossings simulate the way a signal from one neuron to another in the brain can get intensified or dampened according to the strength of the connection. The beams would also pass through simulated neurons that could later regulate their intensities.

Optical computation has been efficient as once light rays are generated, they travel and interact on their own. One can guide them without energy by using a regular glass lenses, whereas transistors would need electricity to operate.

Around 4,000 of those cores on are connected on one single chip by the researchers which would later test its performance with a complex image-recognition task. Well, the computer had to observe people, bicyclists, cars and alternative vehicles within an image and later identify every object properly.

According to the researchers, the chip which is new is not only more efficient than conventional computer chips it even produces way less heat.

Also Read: The US Will Release Chip Shortage Study Details Amid Funding Push