摘要
光学神经网络光电子技术与神经网络模型,有望突破传统电子神经网络计算速度和网络功率效率低的技术瓶颈。为解决教学活动中缺乏实际演示案例问题,设计了利用光学神经网络进行手写数字识别实验。利用光学神经网络芯片级仿真平台Neuroptica构建手写数字识别系统,学生在不同参数下训练模型进行对比实验,也可自主编程设计ONN模型实现其他机器学习任务以验证模型的优劣。
Optical neural network organically combines optoelectronic technology with neural network model.It is expected to break through the technical bottleneck of low computing speed and network power efficiency of traditional electronic neural network.In order to solve the problem of lack of actual case demonstration in teaching activities,an experiment of handwritten digit recognition using optical neural network is designed.The handwritten character recognition system is constructed by using Neuroptica,a chip level simulation platform of optical neural network.Students can train models under different parameters for comparison experiments,and can also independently program and design ONN models to implement different machine learning tasks to verify the performance of the models.
作者
卢瑾
苏豪宇
王艳
任宏亮
LU Jin;SU Haoyu;WANG Yan;REN Hongliang(College of Computer Science&Technology,Zhejiang University of Technology,Hangzhou 310023,China;College of Information Engineering,Zhejiang University of Technology,Hangzhou 310023,China)
出处
《实验室研究与探索》
CAS
北大核心
2023年第11期135-140,共6页
Research and Exploration In Laboratory
基金
浙江省自然科学基金项目(LY20F050009)
上海交通大学区域光纤通信网与新型光通信系统国家重点实验室开放基金(2020GZKF013)。
关键词
光电子技术
光学神经网络
马赫曾德干涉仪
手写数字识别
optoelectronic technology
optical neural network
Mach Zehnder interferometer
handwritten numeral recognition
