Research on Silicon-Based Integrated Diffractive Optical Neural Network
Buch in deiner Nähe kaufen
Beschreibung
This book systematically researches crucial scientific issues such as high integration, high computational capacity, and large-scale expansion of computing units in silicon-based integrated diffractive optical neural networks (DONNs). The contents include theoretical exploration, structural design, simulation calculation, chip fabricating, packaging, testing, and experimental system error compensation for silicon-based integrated DONNs.
The main innovations of this work include the following aspects: (i) A two-dimensional silicon-based spatial electromagnetic propagation model and a phase weight mapping model are proposed, which makes the performance of DONNs pre-trained on computers closer to that of the physical structures; (ii) A silicon-based integrated DONN chip with low latency, low power consumption, high integration, and high computing capacity has been proposed, reaching the theoretical calculation integration degree about 2×103 computing units per square millimeter; （iii） A design method for multifunctional silicon-based integrated DONN chips has been proposed, which has faster computing speed and lower computing energy consumption, further expanding the application scenarios of DONN chips and improving their computing efficiency.
This work has systematically validated the performance of integrated DONNs on the SOI platform, which will promote exciting future developments in integrated optical computing.
This book systematically researches crucial scientific issues such as high integration, high computational capacity, and large-scale expansion of computing units in silicon-based integrated diffractive optical neural networks (DONNs). The contents include theoretical exploration, structural design, simulation calculation, chip fabricating, packaging, testing, and experimental system error compensation for silicon-based integrated DONNs.
The main innovations of this work include the following aspects: (i) A two-dimensional silicon-based spatial electromagnetic propagation model and a phase weight mapping model are proposed, which makes the performance of DONNs pre-trained on computers closer to that of the physical structures; (ii) A silicon-based integrated DONN chip with low latency, low power consumption, high integration, and high computing capacity has been proposed, reaching the theoretical calculation integration degree about 2×103 computing units per square millimeter; （iii） A design method for multifunctional silicon-based integrated DONN chips has been proposed, which has faster computing speed and lower computing energy consumption, further expanding the application scenarios of DONN chips and improving their computing efficiency.
This work has systematically validated the performance of integrated DONNs on the SOI platform, which will promote exciting future developments in integrated optical computing.
Received Outstanding Doctoral Dissertation Award from Tsinghua University Investigates the performance of on-chip diffractive optical neural networks from theory to experimental system Partial results of this dissertation have been published on Nature Communications
Autor*in
Tingzhao Fu
Themen in »Research on Silicon-Based Integrated Diffractive Optical Neural Network«
Diffractive Optical Neural Network Silicon-based Integrated Chip Phase Integration Computing Capacity