And now, a neural network that operates at the speed of light!

Electricity is what usually keeps deep-learning networks running and powering our world with data-fed intelligence. After all, we are talking about computers here.
But chances are that virtual neurons could replace all that copper and wire if small plastic plates, printed using a 3-D printer, can do what some researchers intend them to.
Yes, if a plate starts behaving like a virtual neuron that behaves like its biological counterpart by either transmitting or reflecting incoming light. A recent test has brought the experiment to approximately 95 per cent accuracy. It’s quite a radical concept – if it works on the ground and at a scale. Not just from an angle of applications that hinge a lot on speed but also for reducing energy-usage.

Leave a Reply

Sign Up for NextBigWhat Newsletter