The workshop on mathematics in ML is primarily a networking event to gather relevant Ulysseus researchers, present their current research and discuss potential further steps for collaboration.
Novel conceptual approaches in Machine Learning and Big Data Analytics require increasingly demanding and diverse mathematical apparatus. Statistics and optimization are two mathematical fields that provided theoretical framework for ML from its beginnings. Previous decades witnessed a growing significance of various mathematical theories that have previously been almost invisible in ML. These include Riemannian geometry, Lie group theory, game theory, control theory, information theory and others.
Recent trends emphasize the necessity for systematic and theoretically sound approaches based on mathematical rigor and imagination. Creating the synergy between various research groups working in mathematics and computer science is one of the primary challenges.
The workshop on mathematics in ML is primarily a networking event to gather relevant Ulysseus researchers, present their current research and discuss potential further steps for collaboration.
The main goals may be substantiated in the following three points:
The workshop will be held online.
Interested participants can propose contributions of different kinds. Along with talks on particular mathematical results with (potential) applications in ML, we are also interested in presentations of research groups and broad directions. Finally, talks on mathematical education for ML engineers are also welcome.
The talks are expected to last for 25 minutes.
The event will be concluded by the round table discussion.
Speakers from partner institutions should submit their names, affiliations and titles of talks. Short abstracts (maximum half of the page) are welcome, but not mandatory.
• Optimization in ML
• Probability in ML
• Game Theory in ML. Multi-agent Reinforcement Learning.
• Geometry in ML and Data Science. Geometric Deep Learning.
• Control Theory and ML
• Mathematics for innovative ML solutions and neural network architectures. Geometry informed and physics informed ML.
• Cooperation with industry and potential applications
11:00 – 11:10 | Welcome and Introduction
11:10 – 11:30 | Speaker 1: PhD Cesare Molinari, Postdoc researcher, University of Genova
Topic: Optimization for machine learning
11:30 – 11:50 | Speaker 2: PhD Samuel Vaiter, Research Scientist, Université Côte d’Azur
Topic: Successes and pitfalls of bilevel optimization in machine learning
11:50 – 12:10 | Speaker 3: Hendrik Kleikamp, PhD student, University of Münster
Topic: Certified machine learning for model order reduction of parametrized problems
12:10 – 12:30 | Speaker 4: prof. dr. Vladimir Jaćimović
Topic: Non-Euclidean data and geometric machine learning
12:30 – 13:15 | Round Table Discussion
13:15 – 13.30 | Katarína Valentová, Ulysseus research deputy coordinator, Technical University of Košice
Topic: Presentation on the formation of reserach groups within Ulysseus and other forms of research collaboration
13:30 – 13.45 | Branka Žižić, project manager, University of Montenegro
Topic: Presentation on mobility and project collaborations within Ulysseus
13:45 – 14:00 | Closing Remarks
