Energy-Efficient Neuromorphic Acceleration
This study introduces a 3D heterogeneously integrated electronic-photonic neuromorphic computing architecture that emulates biologically inspired synaptic plasticity and hierarchical learning with exceptional energy efficiency and scalability. The system utilizes optical frequency combs, metaphotonic metalenses, and sophisticated chalcogenide materials to achieve a 1000-fold improvement in synaptic connection and a 100-fold reduction in energy consumption compared to CMOS-based solutions. This platform, integrated with a real-time EPIC simulator, connects neuroscience with next-generation AI technology, facilitating advancements in ultra-efficient brain-inspired computing.
.jpg)
High-Density : GaAs-Enhanced Photonics Integrated Circuits
This research is centered on developing advanced Photonics Integrated Circuits (PICs) using Gallium Arsenide (GaAs), aiming to enhance the density of these circuits by tenfold. By exploiting the unique properties of GaAs, the project seeks to significantly improve the efficiency, speed, and compactness of PICs. This tenfold increase in density has the potential to revolutionize telecommunications and computing, offering a pathway to highly advanced, efficient, and miniaturized optical technologies.
Enhancing Optical Computing Performance through Advanced Fourier Optics Techniques
The innovative application of advanced Fourier optics in optical communications and computing. By leveraging the 4f system, the research aims to achieve high-efficiency optical processing and communication through the development and optimization of Fourier optics convolution, hash encryption techniques, and orbital angular momentum (OAM) multiplexing. The research investigates the integration of post-quantum hashing algorithms with the 4f system for secure and efficient optical hash encryption, as well as the utilization of OAM beams for high-capacity data transmission in free-space optical communications. This research also explores the potential of using metasurfaces for constructing advanced integrated optical neural network acceleration systems, paving the way for groundbreaking advancements in optical computing and machine learning. This pioneering approach has the potential to revolutionize optical communications and computing, unlocking new possibilities for secure solutions .
