Schedule, Information, and Material
This extended short course introduces and discusses basic concepts of machine learning and how they relate to principles and applications of control and optimization theory.
The first 4 weeks will be held given by Timm Faulwasser and team and focus on optimization aspects of the training of neural networks.
The second part is provided by Enrique Zuazua and directed towards control of dynamical systems and how this aligns with the theory of supervised learning.
The last part by Jan Heiland considers applications of neural networks for the formulation of control tasks.
Date | Lecturer | Lecture |
---|---|---|
September 07: 15:00–16:40 | TF/JH | 01 Introduction |
September 09: 15:00–16:40 | TF | 02 Questions and Challenges |
September 14: 15:00–16:40 | TF |
03 Optimization ![]() |
September 16: 15:00–16:40 | TF |
04 Challenges and Tasks in Optimization ![]() |
TF | Note: cancelled | |
September 23: 15:00–16:40 | TF |
05 Gaussian Processes ![]() |
September 28: 15:00–16:40 | TF | ![]() |
September 30: 15:00–16:40 | TF | ![]() |
EZ | Note: cancelled | |
EZ | Note: shifted to Saturday | |
October 09: 15:00–16:40 | EZ |
Note: Saturday ![]() |
October 12: 15:00–16:40 | EZ | ![]() |
October 14: 15:00–16:40 | EZ | ![]() |
October 19: 15:00–16:40 | EZ | ![]() |
October 21: 15:00–16:40 | EZ | ![]() |
October 26: 15:00–16:40 | EZ | ![]() |
October 28: 15:00–16:40 | EZ | ![]() |
November 02: 15:00–16:40 | JH |
17 Physics Informed Neural Network ![]() |
November 04: 15:00–16:40 | JH |
18 Convolutional Neural Networks ![]() |
November 09: 15:00–16:40 | JH |
19 Wrapup and Studying in Germany ![]() |
November 11: 15:00–16:40 | JH |
20 Introduction to PyTorch ![]() |
All times are Beijing time.
All material can be found here. With file names corresponding to the lectures. Please also see the details of every lecture.
Resources
01_Introduction.pdf
01-02-pdes-disc-dynsys.jpg
Resources
01-02-intro-CNN-NSE.pdf
01-02-pdes-disc-dynsys.jpg
Resources
02_ChallengeTasks_Introduction.pdf
03_Optimization_*.pdf
04_ChallengeCasks_Optimization.pdf
04_Solution_ChallengeCasks_Optimization.pdf
05_GaussianProcess*.pdf
Wrapup on how Neural Networks can assist with (or even replace) the costly simulation of PDEs.
Little preview on Data encoding by NN.
What constitutes a Neural Network and how to make it physics informed.
Training an backward mode to compute the gradients for the optimization.
17-0***.jpg
18-0***.jpg
19-01-PINN-ctd.jpg
19-02-PINN-stories.jpg
For all questions around studying abroad, contact the SHU international office.
19-03-study-in-G.jpg
PyTorch
Hello | 你好 |
---|---|
Matrix | 矩阵 |
Vector | 向量 |
Eigenvalue | 特征值 |
Projection | 投影 |
Subspace | 子空间 |
Interpolation | 插值 |
Model | 模型 |
System | 系统 |
Transfer function | 传递函数 |
Stable | 稳定 |
Truncation | 截断 |