Recent Trends in Combinatorial Optimization Augmented Machine Learning - A Graph Learning Perspective

Explore recent developments in combinatorial optimization augmented machine learning (COAML) through this 47-minute conference talk that examines how machine learning and operations research methods integrate to solve complex data-driven problems. Learn about this rapidly growing field that addresses industrial challenges where firms use large, noisy datasets to optimize operations by embedding combinatorial optimization layers into neural networks and training them with decision-aware learning techniques. Discover how COAML excels in contextual and dynamic stochastic optimization problems, including its winning performance in the 2022 EURO-NeurIPS dynamic vehicle routing challenge. Examine recent learning algorithms for empirical cost minimization and structured reinforcement learning, along with new regularizations that exploit connections between local search and Monte Carlo methods. Understand how these algorithms improve performance, reduce computational costs, and lower data requirements while enabling new large-scale applications and providing convergence guarantees that support statistical learning generalization bounds. Follow the unifying theme of graphs throughout the presentation as they play central roles in the problems, architectures, layers, learning algorithms, and theoretical analysis frameworks discussed.