Deep Learning for Science

The course offers knowledge and skills to understand, design, use and manage contemporary solutions of deep learning to handle scientific problems. The course introduces high level tools to use algorithms of deep learning. Students learn usage and working principles of different methods of deep learning. They also study the validation and interpretation of results. Moreover, they learn technological and practical risks that may occur in the application of methods of deep learning.

NB! This course will take place in spring semester 2024/2025 which starts on 3rd of February and ends on 16th of June (you can find that information under Start date section). The real course start and end dates will be announced at the beginning of February at the latest.

After completing this course, the student:

• analyses in which situations to use particular methods of deep learning;
• estimates advantages and disadvantages of methods of deep learning in solution of different problems;
• is able to work with real data;
• chooses and uses suitable algorithms and methods to solve scientific problems;
• estimates content and quality of results of algorithms of deep learning;
• explains main mathematical principles and technological solutions of deep learning;
• independently uses methods of deep learning in solving scientific problems.

Final assessment can consist of one test/assignment or several smaller assignments completed during the whole course. After declaring a course the student can re-sit the exam/assessment once. Assessment can be graded or non-graded. For specific information about the assessment process please get in touch with the contact person of this course. For specific information about grade transfer please contact your home university

• 1. A. Géron, Hands-On Machine Learning with Scikit-Learn & TensorFlow. O’Reilly Media, Inc., 2017.
• 2. M. Erdmann et al, Deep Learning For Physics Research. World Scientific, 2021.

lectures, practices