Scarlet Stadtler
Scarlet Stadtler is a trained meteorologist who focused on atmospheric chemistry during her doctoral thesis. Currently she works as a postdoctoral researcher at the Jülich Supercomputing Centre.
AI methods
KISTE_project
About This Course
This course is designed to teach students about advanced and explainable machine learning methods in environmental science. The course focuses on introducing machine learning methods tailored to earth science and provides students with the skills to explain shallow neural networks, random forest, and UNet. In addition, the course provides an introduction to self-supervised learning and physically consistent transformations for multi-channel remote sensing data. By the end of the course, students will have a solid understanding of advanced machine learning methods and their applications in environmental science. They will have the skills to explain shallow neural networks, random forest, and UNet, and understand the importance of self-supervised learning in machine learning. Students will also have gained experience in using physically consistent transformations for multi-channel remote sensing data, which are essential for accurate analysis of environmental data.
Requirements
- Neural Networks:
- Students should be able to define what a neural network is and explain how it works.
- Students should understand the basic concepts of forward and backward propagation.
- Students should be able to explain how neural networks can be used for classification and regression problems.
- Random Forest:
- Students should be able to define what a random forest is and explain how it works.
- Students should understand how random forests are used for classification and regression problems.
- CNNs:
- Students should be able to define what a convolutional neural network (CNN) is and explain how it works.
- Students should be able to explain how CNNs are used for image classification and object detection tasks.
- Supervised and Self-Supervised Learning:
- Students should understand the concept of labeled and unlabeled data and how they are used in each learning approach.
Lecturers
Timo Stomberg
Timo Stomberg is a trained physicist and is currently working on his doctoral thesis at the University of Bonn within the KISTE project.
Ankit Patnala
Ankit Patnala is a trained computational scientist and is currently working on his doctoral thesis at the University of Bonn and Jülich Supercomputing Centre within the KISTE project.
Further Reading/Watching
- Explainable Machine Learning Reveals Capabilities, Redundancy, and Limitations of a Geospatial Air Quality Benchmark Dataset
- AQ-Bench: a benchmark dataset for machine learning on global air quality metrics
- JUNGLE-NET: USING EXPLAINABLE MACHINE LEARNING TO GAIN NEW INSIGHTS INTO THE APPEARANCE OF WILDERNESS IN SATELLITE IMAGERY