Online Combinatorial

Online combinatorial optimization tackles the challenge of making optimal sequential decisions from a set of possibilities, aiming to maximize overall reward or minimize cost under constraints. Current research focuses on leveraging machine learning, particularly graph neural networks and support vector machines, to improve algorithm performance beyond worst-case scenarios, often incorporating predictions to guide decision-making. These advancements are impacting various fields, including resource allocation (e.g., ad allocation, revenue management), and recommender systems, by enabling more efficient and effective solutions to complex real-world problems. The development of robust and efficient algorithms, particularly those incorporating imperfect predictions, remains a key area of investigation.