On September 3, Microsoft Research in Cambridge, UK, announced that its team had successfully built a new type of Analog Optical Computer (AOC) using Micro LED light sources, optical lenses, and smartphone camera sensors. The research findings were published in the latest issue of *Nature* and are expected to bring groundbreaking applications to financial settlement, medical imaging, and artificial intelligence.
Image source: Microsoft
Unlike digital computers that rely on binary operations, AOC performs calculations through optical physics processes, thus avoiding some of the bottlenecks of traditional digital computing. Researchers envision that AOC could achieve a hundredfold increase in speed and energy efficiency for specific problems, while also possessing advantages such as room-temperature operation and low energy consumption.
It is understood that AOC has already demonstrated its potential in two major areas. In the financial sector, Microsoft has partnered with Barclays Bank to simulate and solve optimization problems in clearing and settling securities transactions. In a test involving 28,000 transactions from 1,800 participants, AOC demonstrated efficient and accurate problem-solving capabilities.
In addition, in the medical field, the Microsoft research team used AOC's "digital twin" model to reconstruct MRI scans, which theoretically could reduce the scan time from 30 minutes to 5 minutes, significantly improving the utilization rate of medical resources.
The research team emphasized that AOC is not a general-purpose computer, but a specialized architecture for optimization problems and specific AI tasks, which uses the intensity changes of photons as they pass through sensors to perform addition and multiplication.
Image source: Microsoft
AOC has expanded its model parameters from 64 to 256. In the future, by adding more Micro LEDs, it is expected to reach millions or even billions of parameters. AOC will be able to perform more complex calculations, and the size of the device will shrink as the device is miniaturized.
In terms of AI applications, Microsoft researchers have demonstrated that AOC can run simple machine learning tasks and shows significant advantages in high-energy-consuming inference tasks such as "state tracking." The research estimates that AOC's energy efficiency is approximately 100 times better than existing GPUs.
Looking ahead, the research team plans to iterate the next generation of AOC every two years, gradually expanding its scale and capabilities. Microsoft's research group believes this technology will play a central role in optimizing financial transactions, accelerating medical imaging, and AI computing.
It's worth noting that Microsoft has been continuously applying Micro LED technology in its latest research projects. Previously, the Microsoft research team also developed the MOSAIC optical interconnect technology based on Micro LED, employing hundreds of low-speed parallel Micro LED channels to achieve a Wide and Slow (WaS) communication architecture with extremely low power consumption, longer transmission distances, higher reliability, and scalability and compatibility. (Compiled)
