Within the thematic program of the Centre de Recherches Mathématiques (CRM) on "Mathematical Foundations of Data Science," we are pleased to announce the Combinatorial Optimization and Data Science workshop, taking place from May 7-9, 2025, in Montreal.
This workshop aims at gathering recent developments on the intersection of Combinatorial Optimization and Data Science, and promote the discussion on emergent future research directions. The three-day workshop will consist of invited talks from distinguished speakers.
What is the optimal fleet size to maximize public transportation service ? What is the optimal route for last mile delivery drones ? Which artificial neural network components can be removed without a significant accuracy loss ? These are all issues with inherent discrete decisions, frequently modeled and tackled with the mixed-integer programming (MIP). MIPs are NP-hard and thus, theoretically intractable. Therefore, the development of heuristics and their use within exact approaches has been extremely important for solution approaches that perform well in practice. In this context, the use of machine learning to improve general purpose MIP solvers or to automatically design greedy heuristics, to name a few, are examples connecting combinatorial optimization with the recent appealing domain of machine learning. To this end, large sets of MIP instances together with their relevant features are necessary, linking the broader domain of Data Science. Moreover, the inclusion of machine learning forecasts within decision making becomes more and more crucial to more suitably reflect real-world problems. On the other side, the use of mixed-integer programming to sparsify or build adversarial attacks to neural networks, to model new regression trees, or to optimize feature selection, among others, illustrate well the synergies of Combinatorial Optimization and Data Science.
The workshop Combinatorial Optimization and Data Science aims at gathering recent developments on the intersection of these domains and promote the discussion on emergent future research directions. A variety of real-world decision problems and the role of data science to tackle them will be highlighted through a series of talks from expert researchers in this field. Likewise, the role of combinatorial optimization within machine learning methodologies will be surveyed within the workshop presentations.
Registration is now open.
More information is available here: https://www.crmath.ca/en/activities/#/type/activity/id/3992
We look forward to seeing you at the workshop!
Organizing Committee:
Quentin Cappart
Margarida Carvalho
Utsav Sadana