A new processor that could revolutionize artificial intelligence has just been proposed in Nature in an international team of researcher, including professor Roberto Morandotti of the Institut National de la Science Scientifique (INRS).
Next-generation artificial neural networks, layers of entangled artificial neurons, are becoming increasingly popular for machine learning tasks. Currently, computers are reaching their functional limits, yet the need for ever more advanced technologies is increasing by the day.
Photons were used instead of electrons, to bring information through fiberoptics. Currently, not only electrons can conduct certain functions at the speed of light, but they are the backbone of the modern Internet, where it is important to prevent electronic bottlenecks (conversion of an optical signal into an electronic signal, and vice versa).
The proposed optical neural network is able to process and recognize large-scale data and images at faster processing speeds than are feasible with traditional electronic computers. Professor Morandotti, a specialist in integrated photonics, instructs us on how a light source composed of several evenly scattered light waves could power an optical computing chip.
This device provides a way of doing a cross-correlation for image processing. The paper offers positive findings for real-time massive-data machine learning functions, such as recognizing faces in cameras and realizing how cells function. Their methodology can be scaled to work alongside other networks with diverse technologies such as autonomous vehicles and real-time video detection allowing for a potential convergence with the evolving Internet of Things.