课程号：TE206004

课程名：深度学习（国合部海外教师引课）

英文课程名：Deep Learning

开课单位：通信工程学院

授课教师：Amir Sadovnik

责任教师：肖潇   肖潇

学分/学时：1学分/16学时

教师简介：Amir Sadovnik是田纳西大学EECS系的副教授。他在康奈尔大学电子与计算机工程学院获得博士学位，并得到高级多媒体处理实验室成员陈辰教授的指导。在来到康奈尔大学之前，他在库珀联盟获得电子与计算机工程学士学位。在来到UT之前，Amir是宾夕法尼亚州伊斯顿市拉法叶学院的助理教授。他花了四年时间，除了从事本科生研究工作外，大部分时间教授本科生水平的课程。在拉斐特期间，他教授了各种入门和高级计算机科学课程。此外，他帮助重新设计了计算机科学入门课程，使其更具包容性，并积极倡导该领域的应用。他在计算机视觉领域的研究主要是由人类理解和与图像交互的方式驱动的。这种以人为中心的观点使他从事新的、令人兴奋的项目，这些项目利用来自不同领域的工具(如计算机视觉、信号处理、自然语言处理、机器学习等)，并以新的方式应用它们。他目前的研究主要集中在使用深度神经网络来完成一些更主观的任务，比如唤起情感、相似的面孔和时尚的兼容性。Engineering at Cornell University and was advised by Prof. Tsuhan Chen as member of the Advanced Multimedia Processing Lab. Prior to arriving at Cornell he received his Bachelors in Electrical and Computer Engineering from The Cooper Union.Prior to arriving at UT, Amir was an assistant professor at Lafayette College in Easton, PA. He spent four years mostly teaching undergraduate level courses in addition to working on undergraduate research. During his time at Lafayette he taught a variety of both introductory and advanced computer science courses. In addition, he helped redesign the introductory computer science course to make it more inclusive and was an active advocate for women in the field.His research in the field of computer vision has been mostly driven by the way humans understand and interact with images. This human centered view has led him to work on new and exciting projects, which utilize tools from different fields (such as computer vision, signal processing, natural language processing, machine learning, etc.) and apply them in new ways.His current research is mostly centered on using deep neural networks for tasks which tend to be more subjective such as evoked emotions, face similarity, and fashion compatibility. The subjective nature of these problems presents many interesting obstacles and opportunities which he explores in his research.

课程简介：如何建立表示高维数据的深度网络的理论和实践方面。我们将讨论监督学习和非监督学习的深度模型，包括卷积神经网络、自编码器、生成对抗网络和递归神经网络。在本课程结束时，学生将实现以下目标:1、了解人工神经网络包括深度神经网络的基本特性。2、了解如何训练深度神经网络被训练3、了解深度学习可能的不同架构以及它们适合的应用程序。4、了解如何实际处理不同的网络参数，以达到最优的结果。5、将深度学习技术应用于新的应用。AT THE CONCLUSION OF THIS COURSE STUDENTS WILL ACHIEVE THE FOLLOWING:1、Understand the fundumentals of artifical neural networks including deep neural networks.2、Understand how to train deep neural networks are trained.3、Understand the different architectures of deep learning possible and the applications they are suitable for.4、Understand how to practically deal with the different network parameters in order to achieve optimal results.5、Apply deep learning techniques to novel applications.

先修课程：信号与系统