本所声明  |  联系方式  |  中国科学院  |  数字认证(OA)   |  ARP  |  English  |  邮箱

8月31日学术报告:基于神经网络的深度学习模型及其应用

来源: 发布时间:2018-08-30【字体:

/Topic基于神经网络的深度学习模型及其应用,

             Deep Learning Model Based on Neural Network and its Applications 

时间/Time 831   10:00AM 

地点/Place: 多功能厅,SIOM, Shanghai  

报告人/Speaker:  叶世伟,Ye Shiwei 

    

  本报告从神经网络的介绍开始,然后简单介绍目前主要的两类基于神经网络的深度学习模型:一类是前馈类型的网络,另外一类为反馈类型的网络。前馈类型的网络主要包括前馈性的多层前馈网络、深度卷积网络、残差网络模型、高速网络模型和密集网络模型.反馈类型的网络主要包括简单递归神经网络和含有长短期记忆的循环网络。最后讨论根据深度神经网络应用中存在的问题,提出改进方法,主要包括胶囊网络与应用、生成对抗网络和基于物理模型的深度网络结构设计方法。     

  This report begins with the introduction of neural networks, and then briefly introduces two main types of deep learning models based on neural networks: one is feedforward network, the other is feedback network. Feedforward-type networks mainly include multi-layer feedforward networks, deep convolution network, residual network, high-way network and dense network. Feedback-type networks mainly include simple recursive neural networks and long short-term memory neural network. Finally, according to the existing problems in the application of deep neural network, the improvement methods are proposed, including capsule network and its application, generative adversarial networks and physical model-based deep network structure design method. 

    

  叶世伟,1968年出生,中国科学院大学电子电气与通信工程学院副教授。 1995年在中国科学院计算技术研究所获得计算机科学理论专业博士学位。 主要从事智能信息处理,优化理论和算法设计,已经发表论文40余篇。 

  Ye Shiwei, born in 1968, is an associate professor at the School of Electrical, Electronic and Communication Engineering, University of Chinese Academy of Sciences. In 1995, he obtained a doctorate in computer science theory from the Institute of Computing Technology of the Chinese Academy of Sciences. He is mainly engaged in intelligent information processing, optimization theory and algorithm design, and has published more than 40 papers. 


附件下载: