Scalable, Energy-Efficient Optical-Neural Architecture for Multiplexed Deepfake Video Detection
Researchers have developed a novel hybrid digital-analog system for detecting deepfake videos, leveraging optical computation for efficient, high-throughput inference. This architecture combines a lightweight digital front-end with a spatially multiplexed optical decoder, enabling the simultaneous processing of multiple video streams in a single pass. The system demonstrated high accuracy in detecting various types of deepfakes, including AI-generated content, while also showing resilience against noise, compression, and adversarial attacks. AI
IMPACT This hybrid optical-AI approach could significantly reduce the computational cost and energy consumption of deepfake detection, enabling more scalable and robust real-time monitoring systems.