Artificial Intelligence (AI) is the most important modern technique in the scientific study since the adoption of quantum mechanics and provides promising findings in a wide variety of fields of science and technology. The international team of researchers, including Professor Roberto Morandotti of the Institut National de la Recherche Scientifique (INRS), has just unveiled a new photonic processor that could revolutionize artificial intelligence, as stated by the prestigious journal Nature.
An international team of researchers introduced a new photonic processor that could revolutionize artificial intelligence. Artificial neural networks, layers of interconnected artificial neurons, are of great interest in machine learning tasks such as speech recognition and medical diagnosis.
Lightmatter is a startup that is trying to improve the way all the programming is done—and not in a small way. Artificial neural networks, layers of entangled artificial neurons, are of significant importance to machine learning activities such as speech recognition and medical diagnosis. In fact, electronic computer hardware is approaching the limits of its capability, and the need for greater computing capacity is gradually rising.
Researchers switched to photons instead of electrons to bring information at the speed of light. In fact, not only can photons process information much faster than electrons, but they are the backbone of the modern Internet, where it is necessary to escape the so-called electronic bottleneck (conversion of an optical signal into an electronic signal, and vice versa).
Increased Computing Speed
The proposed optical neural network is capable of detecting and manipulating large-scale data and images at ultra-high computational speeds of more than 10 trillion operations per second. Professor Morandotti, a pioneer in integrated photonics, discusses how an optical frequency comb, a light source with several evenly spaced frequency modes, was integrated into a computer chip and used as an effective source for optical computation. At the base of all that AI and machine learning is, like most computing operations, a lot of math (hence the name computing). A general-purpose computer can do any of that math, but for complex problems, it has to break it down into a series of smaller ones and perform them sequentially.
This system performs a method of matrix-vector multiplication known as an image processing convolution. It shows positive results for real-time massive data machine learning activities, such as recognition of faces in cameras or identification of pathologies in clinical scanning applications. Their solution is flexible and adaptable to far more complex networks for challenging applications such as autonomous vehicles and real-time video detection, enabling complete convergence of the up-and-coming Internet of Things in the not-so-distant future.
The challenge is that as demand grows for AI-based goods, these measurements need to be completed more and more rapidly, but we are approaching the limits of how quickly and effectively they can be rendered and relayed back to the customer. So while the computing infrastructure that has existed for decades is not going anywhere, there are promising options on the horizon for certain niches.