Track Editors

Christian Bessiere, CNRS, University of Montpellier
Eugene Freuder, University College Cork
Tias Guns, KU Leuven
Lars Kotthoff, University of Wyoming
Michela Milano, University of Bologna
Gilles Pesant, Polytechnique Montréal

Overview

The objective of this JAIR special track is to encourage and showcase work combining Constraint Programming (CP) and Machine Learning (ML). Constraint Programming is taken to encompass all forms of constraint satisfaction and optimization, as exemplified by the Constraint Programming conference and the Constraints journal. Machine Learning is taken to include both classical and neural network approaches. The combination may involve applying ML to CP, CP to ML, or both working together towards some objective or application.

Considerable interest has been shown in combining CP and ML, with special tracks at Constraint Programming conferences and Bridge events at the AAAI conference. Applying machine learning to automate the modelling and solving of constraint programming has been a central component of the longstanding “Holy Grail” of CP, which has been pursued with a series of workshops. A Machine Learning for Constraint Programming summer school was held recently. Combining CP with ML fits into the broader objective of effectively bringing together contributions from different AI subfields, and more generally “symbolic” and “neural” AI.

Topics for this special track include, but are not limitied to:

• Acquiring constraints
• Learning heuristics
• Performance optimization through tuning
• Algorithm selection
• Learning constraint models
• Using LLMs/Chatbots to solve constraint satisfaction problems
• Using CP to address ML limitations
• ML for combinatorial optimization
• Learning preferences
• Explanation/reformulation of statistical models with constraints
• Embedding constraint reasoning in ML
• Combining prediction and optimization
• CP for fairness in ML
• Synthesizing CP models from natural language

Status

The track is closed for new submissions. Accepted submissions will be added to this page on publication.