[WiSe 2025/26] Artificial Intelligence in Human-Water Systems

Learning Outcomes

In this block course, students will learn about common machine learning (ML) methods and concepts in the field of human-water systems. The focus of the course is mainly on the application of models and methodologies for geospatial and spatial-temporal analysis, while fundamental principles are covered briefly at the beginning of each session. In addition, the students will learn the fundamentals of deep learning (DL) through examples of recent DL techniques and their application in the field of human-water systems. Since the course focuses mainly on the actual application of Artificial Intelligence (AI), it will also include interactive sessions about principles of software development comprising basic and advanced usage of version control like automated testing via continuous integration. The block course is split into two parts: the first part covers daily sessions and hand-on exercises over the course of two weeks, followed by a second part comprising the development of a final AI programming project and the preparation of a related written report. At the end of the in-person sessions, the students will have acquired a broad understanding of common ML and DL models and techniques, providing them to further deepen their knowledge in the field of AI through a final programming project (second part of the course). Through this project the students will develop advanced knowledge about a specific model and application field of their own interest. They summarize their findings in a related written report. Upon completion of the entire course, the students will be able to conduct their own AI project including (a) data preparation, (b) method selection, (c) training and (d) evaluation.
 

Content

In recent years, with increasing data volumes, the use of AI methods in environmental applications has revealed new insights. This course addresses this topic by focusing on the complex geospatial relationships between humans and hydrological systems. Examples include AI-based risk assessments of natural hazards, flood detection using remote sensing data, forecast of hydrological extreme events or analysis of text-based data. Topics covered in the course (including but not limited to): - Common classification models - Regularized linear regression - Clustering algorithms - Ensemble learners - Bayesian Networks as an example of probabilistic, explainable AI - Introduction to Neural Networks and Deep Learning - Interactive sessions on the use of version control systems for efficient collaborative (multi-person) software development Additional topics: - Model performance evaluation and error metrics - Hyperparameter tuning - Overfitting and regularization - Feature selection - Examples for efficient (geospatial) data handling and model testing are given in the lab.
 

Description of Teaching and Learning Methods

The block course consists of a lecture and a hands-on lab part, which are closely coupled. At the end of the course the students will submit an AI project using a machine or deep learning model of their choice and a short report. The students can form groups of up to three students to conduct the final project and report, but this is optional. In case of group work, it is expected that the group will apply a higher number of AI-related methods and techniques to solve the project. The written reports should comprise at least following number of words: • in case a project is conducted alone: minimum 1500 words • in case of a group of two students: minimum 2500 words • in case of a group of three students: minimum 3500 words In the last session of the block course, the students will pitch their project idea. This 15-minutes presentation should include an overview of their project, a short state-of-the-art section, overview of the model(s) and techniques they plan to use, and expected challenges. Instructions about the final schedule, lecture rooms, and on how to get access to the course and exercise materials will be communicated to the registered students via the e-learning ISIS platform.
 
 

Desirable prerequisites for participation in the course:

Students from all study programs are invited to join the course. It is pre-required that students have at least basic to intermediate programming skills in Python and are familiar with tools for package management (at least Anaconda or Mamba). Basic knowledge in version control systems like GitLab or GitHub is beneficial but not mandatory. It is desirable, yet not required, that interested students have solid knowledge of common statistics and data analysis techniques.