Martin Krutský is a ML/AI researcher at the Czech Technical University where he is currently working towards his PhD. On the technical side of his research, he is interested in explainability techniques for deep neural networks, especially graph neural nets and other structured models. His previous work is in the field of geometric deep learning and symmetries in machine learning. On the hand, he is also interested in the intersection of AI research with other fields, especially concerning a responsible and ethical use of AI and AI safety.
Graduate
Czech Technical University: M.S. in Artificial Intelligence
Undergraduate
Czech Technical University: Computer Science and Artificial Intelligence
Representative Publications
- Martin, K. (2021). Basics of symmetries in deep learning (Bachelor’s thesis, Czech University of Technology in Prague. Computing and Information Center.).
- Martin, K. (2024). Interpretable deep learning with symmetries for planning (Master’s thesis, Czech University of Technology in Prague. Computing and Information Center.).
Deep Learning
Explainable AI
Geometric Deep Learning
ML Engineering
Your ideal research study
Robust explanation of global concepts in large transformer models
What drew you to consciousness
The question of how similar do artificial neural networks behave compared to the biological ones, despite the difference in substance and processes.
Hobbies
chess, badminton, functional training, specialty coffee