| 课程号 |
00333734 |
学分 |
3 |
| 英文名称 |
Data-Driven Optimization and Learning |
| 先修课程 |
无 |
| 中文简介 |
本课程广泛介绍了使用计算机仿真和流数据分析和优化动态随机模型的重要方面。重点是持续优化和学习及其广泛的应用领域：从社交网络到计算机网络，从金融工程到业务流程。该课程将向学生介绍递归算法的使用，通过基于仿真/数据驱动的方法来分析动态随机模型，以进行优化和学习。本课程的主要问题是如何使用仿真/流数据为现实生活中的问题做出更好、更负责任的决策。该课程还将反思我们在社会中见证的技术和数学发展。 |
| 英文简介 |
This course gives a broad treatment of the important aspects of the use of computer simulation and of streaming data for the analysis and optimization of dynamic stochastic models. The emphasis is on continuous optimization and learning (i.e., we do not cover discrete optimization in this course). Applications will stem from a wide range of domains: from Social Networks to Computer Networks, and Financial Engineering to Business Processes. The course will introduce students to the use of recursive algorithms in analyzing dynamic stochastic models through simulation-based/data-driven methods for optimization and learning. The leading question of the course is how to use simulation/streaming-data to make better and more responsible decisions for real-life problems. The course will also reflect on the technological and mathematical developments we witness in our societies. While actively working on simulation projects, the course will provide space for reflecting on the mathematical/technological paradigm. That is, next to learning the actual techniques, students will be stimulated to reflect on the history of science and the technological developments around them. |
| 开课院系 |
工学院 |
| 成绩记载方式 |
|
| 通识课所属系列 |
|
| 授课语言 |
英文 |
| 教材 |
|
| 参考书 |
|
| 教学大纲 |
This course gives a broad treatment of the important aspects of the use of computer simulation and of streaming data for the analysis and optimization of dynamic stochastic models. The emphasis is on continuous optimization and learning (i.e., we do not cover discrete optimization in this course). Applications will stem from a wide range of domains: from Social Networks to Computer Networks, and Financial Engineering to Business Processes. The course will introduce students to the use of recursive algorithms in analyzing dynamic stochastic models through simulation-based/data-driven methods for optimization and learning. The leading question of the course is how to use simulation/streaming-data to make better and more responsible decisions for real-life problems. The course will also reflect on the technological and mathematical developments we witness in our societies. While actively working on simulation projects, the course will provide space for reflecting on the mathematical/technological paradigm. That is, next to learning the actual techniques, students will be stimulated to reflect on the history of science and the technological developments around them.
Topics
1. Programming language is Python (basic programs will be provided). Other programming languages, such as Matlab, are also fine but are not supported.
2. Basics of Monte Carlo Simulation: random number generation, output analysis
3. Standard simulation models: queuing systems, social networks, financial products.
4. Data and simulation: combining simulation with available historical data
5. Estimation of gradients via simulation and their application in learning and optimization: stochastic gradient method, stochastic approximation, supervised learning, non-supervised learning
课堂教授
Presentation and written report 30%
Simulation project written report 30%
Final exam 30%
Attendance and discussion 10%
Total 100%
|
| 教学评估 |
|
|