This practical course provides hands-on experience in applying deep learning methods in diverse applications. Our focus this term are medical applications. The course focuses on the development, training, evaluation, and interpretation of deep neural networks under realistic constraints, such as limited data availability and restricted computational resources.

Tasks and Goals
During the practical course, students will work on two project-based tasks, each addressing a specific medical deep learning application. The course covers, among other topics:

-Learning with limited data and computational resources

-Computer vision and medical image processing

-Biosignal and time-series data analysis

-Implementation and training of deep neural networks

-Use of deep learning frameworks such as PyTorch and TensorFlow

For the assigned tasks, students are expected to:

-Design and implement a generic and reusable deep learning framework, including efficient data loading, training, and evaluation pipelines
-Apply regularization techniques to improve generalization
-Perform hyperparameter optimization using grid search or Bayesian optimization
-Use visualization methods to gain insights into trained models
-Track experiments and results using tools such as TensorBoard and Weights & Biases
-Use Git for version control and collaborative development

Expectations and Assessment
Students will work in groups of two. Each group is required to complete both projects:

One project will be presented as a poster presentation
The other project will be documented in a written report
Tutorials will be held every week to support the projects and discuss relevant methods.