Customize Consent Preferences

We use cookies to help you navigate efficiently and perform certain functions. You will find detailed information about all cookies under each consent category below.

The cookies that are categorized as "Necessary" are stored on your browser as they are essential for enabling the basic functionalities of the site. ... 

Always Active

Necessary cookies are required to enable the basic features of this site, such as providing secure log-in or adjusting your consent preferences. These cookies do not store any personally identifiable data.

No cookies to display.

Functional cookies help perform certain functionalities like sharing the content of the website on social media platforms, collecting feedback, and other third-party features.

No cookies to display.

Analytical cookies are used to understand how visitors interact with the website. These cookies help provide information on metrics such as the number of visitors, bounce rate, traffic source, etc.

No cookies to display.

Performance cookies are used to understand and analyze the key performance indexes of the website which helps in delivering a better user experience for the visitors.

No cookies to display.

Advertisement cookies are used to provide visitors with customized advertisements based on the pages you visited previously and to analyze the effectiveness of the ad campaigns.

No cookies to display.

Ad image

Light-based chips could help quench AI’s growing thirst for energy

MONews
4 Min Read

“What we have here is something incredibly simple,” he said. Wu Tianwei, lead author of the study. “We can change and reprogram the laser pattern on the fly.” Researchers used this system to design a neural network that successfully distinguished vowel sounds. Most photonic systems require training before they can be built. Because education inevitably involves reorganizing connections. But because the system can be easily reconfigured, the researchers mounted it on a semiconductor and then trained the model. Now they plan to increase the amount of data it can process by increasing the size of the chip and encoding more information in different colors of light.

It’s a development that even Psaltis, which built a facial recognition system in the 90s, finds impressive. “The wildest dreams we had 40 years ago were tiny compared to what actually happened.”

first rays

Although optical computing has advanced rapidly over the past few years, it is still a long way from replacing electronic chips that run neural networks outside the lab. The paper presents photonic systems that perform better than electronic systems, but typically run small-scale models using older network designs and smaller workloads. And many of the reported numbers for photon supremacy don’t tell the whole story, said Bhavin Shastri of Queen’s University in Ontario. “It’s very difficult to do a one-to-one comparison with electronics,” he said. “For example, when you use a laser, you don’t really talk about the energy that powers the laser.”

Laboratory systems must scale up before they can demonstrate competitive advantage. “How big do you have to be to win?” McMahon asked. Answer: It is unusually large. That’s why no one can match the chips made by Nvidia, the chips that power today’s most advanced AI systems. There’s a huge list of engineering puzzles to figure out along the way. This is a problem that the electronics field has been solving for decades. “Electronics is starting out with a big advantage,” McMahon said.

Some researchers believe that ONN-based AI systems will first succeed in specialized applications where they offer unique advantages. Shastri said one promising use is countering interference between various wireless transmissions, such as 5G cellular towers and radar altimeters that help airplane navigation. Earlier this year, Shastri and several colleagues ONN was created It can classify multiple transmissions and select signals of interest in real time, with processing latency of less than 15 picoseconds (15 trillionths of a second). This is less than 1/1000th of the time that electronic systems take and uses less time. than 1/70th of the force.

But McMahon said the grand vision of optical neural networks that could outperform electronic systems for general use is still worth pursuing. his group last year I ran the simulation. We show that within a decade, sufficiently large optical systems could make some AI models more than 1,000 times more efficient than future electronic systems. “Many companies are currently working hard to achieve 1.5x profits. “It would be truly amazing to see thousands of times the profits,” he said. “If successful, it will probably be a 10-year project.”


original story Reprinted with permission. Quanta Magazine, Editorially independent publication Simmons Foundation Our mission is to enhance public understanding of science by covering research developments and trends in mathematics, physical sciences, and life sciences.

Share This Article