| 课程号 |
04833760 |
学分 |
3 |
| 英文名称 |
DeeCamp AI Training |
| 先修课程 |
数据结构、概率统计 |
| 中文简介 |
本课程面向全球计算机、电子工程、数学相关专业的博士、硕士及本科在校生，旨在培养应用型高等AI人才。课程开课期为4周，包含2阶段的学习与实践：
第一部分，7月23日-7月29日，介绍深度学习、人工智能的基本原理、方法以及在不同领域的实践，包括深度学习的发展
现状、理论以及基本方法；介绍深度学习框架TensorFlow；以及深度学习在不同领域的应用包括计算机视觉、自然语言理解等。
第二部分，7月30日-8月23日，学生以组队的形式与AI工程指导老师一起，参与实际的AI项目开发与设计。 |
| 英文简介 |
The course aims to give undergraduate, graduate, and Ph.D students in computer related disciplines an overview of the AI related technology, and take them into the practice of developing AI applications in 4 weeks. It is consisted of two parts:
The first part goes through July 23-July 29, and introduces the knowledge and application of deep learning, natural language processing, computer vision, and Tensor Flow, and etc.
The second part goes through July 30-August 23. Students will form teams and develop AI applications based on real subjects, under the mentoring of both industry experts and university professors. |
| 开课院系 |
信息科学技术学院 |
| 通选课领域 |
|
| 是否属于艺术与美育 |
否 |
| 平台课性质 |
|
| 平台课类型 |
|
| 授课语言 |
中英双语 |
| 教材 |
Deep Learning,Ian Goodfellow, Yoshua Bengio and Aaron Courville,MIT Press,2016, |
| 参考书 |
1；
|
| 教学大纲 |
无
Session 1：Infrastructures From Big Data to Deep Learning
- A Full-stack Overview of AI Infrastructures
- Virtualization, Container and Container-management System
- Batch Processing and MapReduce Hell
- Flume – A Pipeline Optimization Framework
- Large-scale Incremental Processing
- Distributed Machine Learning Frameworks
- Deep Learning – Infrastructure Considerations
- Machine Learning Visualizations
Session 2: Introduction and Case Studies of Natural Language Processing
- Introduction
- Case studies
- Search engine
- Sentiment analysis
- Chatbot
Session 3: Low-Level Computer Vision
- What is computer vision and what Computer Sees?
- What is color and how color works?
- BRDF, diffuse reflection and specular reflection
- Channel, pixel and HDR
- What and how computer sees?
- Detecting, encoding, image stich and depth of field
- 2D Projection of the 3D World
- Homogeneous coordinates
- Projection = Perspective Matrix
- SfM: Structure from Motion
- Objective Function
Session 4: Introduction and Application of Tensor Flow
- What is Tensor Flow?
- Trending in Tensor Flow and other Deep Learning Frameworks
- Data Flow Model
- Programming Model
- Underlying Implementations
- Extensions
- Gradient Computation
- Partial Execution
- Multi-GPU in Single Machine and Distributed Training
- APIs on FCN, CNN, RNN
Session 5: Case Studies of Machine Translation
- Case Studies of Machine Translation
- Frameworks
- Technology Applications
- Challengers
Session 6: Case Studies of Object Recognition with Computer Vision
- Case Studies of Object Recognition with Computer Vision
- Frameworks
- Technology Applications
- Challengers
Session 7: Case Studies of Autonomous Driving
- Case Studies of Autonomous Driving
- Frameworks
- Technology Applications
- Challengers
第一阶段教学方式以教师ppt授课为主；
第二阶段以学生分组实操为主，配合业界老师直接进行指导。
同时借助多媒体设备展示相关的动画、视频等内容。
学生成绩由三部分构成，课程参与占30%，课程项目大作业占40%，调研报告及展示占30%。
|
| 教学评估 |
|
|