The course machine learning and its application in the communications technology provides an overview of modern machine learning, the fundamentals (e.g., neural network architecture, loss function, backward propagation, optimizer and training), advanced neural network architectures (convolutional neural network, transformers, recurrent neural network, graph neural network), and case studies in the communications technology. Purpose of this course is both competence in the job market as machine learning talent and ability for innovation in the next-generation communication systems.
After the lecture, the students
- know the basics of neural network models
- understand the training process with massive data for supervised learning
- can generalize from supervised learning to unsupervised learning
- can implement and train the neural network model with Python and Pytorch for simple tasks
- understand how to consider domain knowledge of communications engineering in designing the neural network architecture and objective
- can optimize the training process if the outcome is not as expected.
Contents of the lecture are
- Introduction of basic ideas of neural networks
- Introduction of the basic neural network architecture as well as loss function, gradient descent and optimizer for neural network training
- Setting up a development environment for machine learning with Python and Pytorch
- Hands-on experiment of defining and training of a simple deep neural network
- Introduction of advanced neural network architectures, including convolutional neural network, recurrent neural network, graph neural network and transformers. Understanding why they were invented and how they work
- Introduction of dedicated objective function for unsupervised learning in communications engineering
- Introduction of dedicated neural network architectures for unsupervised learning in communications engineering
Time and location: every summer semester, Thursday, 13:15 - 14:45 (for both lecture and exercise), starting from April 13th 2023, SN22.2, Schleinitzstraße 22.
Lecturer: Dr. Bile Peng
Assistent: Ramprasad Raghunath, M.Sc.
Credit points: 6
Examination: oral or written exam