Organization¶
Lecturers¶
Sebastian Egli |
Boris Thies |
|---|---|
|
|
sebastian.egli@geo.uni-marburg.de |
boris.thies@geo.uni-marburg.de |
Grading¶
6 ECTS credits are awarded for successfully completing this course.
Three case studies and three corresponding challenges build the basis of this course.
Studienleistung:
The course credit consists of the submission of a well-prepared notebook for publication on the course website for one of the three case studies covered.
Students are assigned to case studies at the beginning of the semester.
In addition, at least one result must be submitted for each challenge before its respective deadline.
Prüfungsleistung:
The course grade is determined by the evaluation of three well-prepared notebooks, one for each challenge.
The deadline for the submission of these three notebooks will be communicated during the semester.
Prediction results of the models must be reproducible, so please do always set a random state during model training!
Evaluation criteria are:
Selection of the ML model: Is the ML model suited to solve the problem stated in the challenge?
Programming approach: Has a suitable programming approach been selected?
Layout: Is the notebook clearly arranged? Does the layout follow a clear and understandable concept?
Code quality: Have redundant code sections been removed from the notebook?
Documentation: Has the code been fully and comprehensibly documented?
Presentation of training/validation/testing procedures: Are the work packages adequately presented
Visualization: Are intermediate steps (where useful for understanding) and model outputs clearly visualized? (title, axis labels, plot size, etc.)
Use of Python libraries: Have the Python libraries from the course been used sensibly? (pandas, xarray, matplotlib, sklearn, tensorflow, …)
Software requirements & setup¶
Install Miniconda on your system. Installation instructions can be found here.
Download this environment.yml file and use it to create a new conda environment:
conda env create -f environment.yml
This installs all packages in a conda environment called “ML_course”, which we will be using throughout this course.
Activate the environment via:
conda activate ML_course
Now all system variables should be set up correctly and we are good to go.
Most of the time, we will be working with JupyterLab. Start it via:
jupyter lab
Literature¶
The following sources give interesting overviews and insights into the topics covered in this course:
Ma, L., Liu, Y., Zhang, X., Ye, Y., Yin, G., & Johnson, B. A. (2019). Deep learning in remote sensing applications: A meta-analysis and review. ISPRS Journal of Photogrammetry and Remote Sensing, 152(March), 166–177. [MLZ+19]
Maxwell, A. E., Warner, T. A., & Fang, F. (2018). Implementation of machine-learning classification in remote sensing: an applied review. International Journal of Remote Sensing, 39(9), 2784–2817. [MWF18]