Advancing Intelligent Neuromorphic Computing: Recent Progress in All-Optical-Controlled Artificial Synaptic Devices
Jian Yao, Yu Teng, Qinan Wang, Yuqi He, Liwei Liu, Chun Zhao, Lixing Kang
Abstract
The rapid development of artificial intelligence and the increasing volume of generated data have heightened the demand for computational power. However, the traditional von Neumann architecture encounters performance bottlenecks due to frequent data transfers and high energy consumption. A promising solution is integrating functions such as perception, storage, and processing into a single device, known as neuromorphic devices. Currently, most neuromorphic devices rely on fully electronic or electro-optic hybrid control, which limits their speed and energy efficiency. In contrast, all-optical-controlled neuromorphic devices provide faster data transmission, lower energy consumption, and better scalability. This review analyzes the latest advancements in all-optical-controlled neuromorphic devices, with a particular focus on the exploration of materials. It also presents a detailed analysis of the physical mechanisms that underpin all-optical-controlled neuromorphic computing, offering insights into the fundamental operation of these devices. Unlike previous reviews, which primarily focus on the general characteristics of neuromorphic devices, this work examines the contributions of materials and all-optical-controlled mechanisms in improving efficiency and scalability. Additionally, the diverse applications of all-optical-controlled neuromorphic devices in optical logic gates, visual perception, and brain-inspired computing are discussed, illustrating their potential to influence computational paradigms.