Light-powered smart system can read complex light patterns
A new system developed by researchers at Wits University can detect hidden information within light without requiring detailed image processing.
Scientists at Wits University have developed a light-driven artificial intelligence system capable of recognizing hidden information in complex light patterns, promising to accelerate future communication systems.
Background of the Research
Researchers at Wits University combined their expertise in artificial intelligence and physics to unveil a new light-based system.
This developed method allows information to be read using light's own structure without the need for detailed images or intensive computer processing.
Study Details and Team
Published in the journal Nature Communications, this groundbreaking research was designed and led by Dr. Isaac Nape and Professor Andrew Forbes.
Master's student Hadrian Bezuidenhout led the project, with contributions from Cade Peters and Ram Nandan Kumar.
Working Mechanism of the System
The developed system directs light to detectors by passing it through specially designed layers.
In this way, even if distortions occur during transmission, successful recognition of hidden features such as topology becomes possible.
Test Results and Success Rate
Trained on just 15 samples, the light-driven brain successfully identified a total of 81 different light structures.
The system proved that it continues to operate stably even in scenarios where noise is added to the detector.