Generalization

Generalization #

Generalization is a critical aspect of machine learning for combinatorial optimization. This section covers approaches to improve generalization across different problem instances and scales.

Recent Literature #

1. It’s Not What Machines Can Learn It’s What We Cannot Teach ICML, 2020. paper

   Gal Yehuda, Moshe Gabel and Assaf Schuster

2. Learning TSP Requires Rethinking Generalization CP, 2021. paper, code

   Chaitanya K. Joshi, Quentin Cappart, Louis-Martin Rousseau and Thomas Laurent

3. Generalization of Neural Combinatorial Solvers Through the Lens of Adversarial Robustness ICLR, 2022. paper

   Simon Geisler, Johanna Sommer, Jan Schuchardt, Aleksandar Bojchevski and Stephan Günnemann

4. Learning for Robust Combinatorial Optimization: Algorithm and Application INFOCOM, 2022. journal

   Shao, Zhihui and Yang, Jianyi and Shen, Cong and Ren, Shaolei

5. ⭐ROCO: A General Framework for Evaluating Robustness of Combinatorial Optimization Solvers on Graphs ICLR, 2023. paper, code

   Lu, Han and Li, Zenan and Wang, Runzhong and Ren, Qibing and Li, Xijun and Yuan, Mingxuan and Zeng, Jia and Yang, Xiaokang and Yan, Junchi

6. Towards Omni-generalizable Neural Methods for Vehicle Routing Problems ICML, 2023. paper, code

   Zhou Jianan, Yaoxin Wu, Wen Song, Zhiguang Cao, Jie Zhang

7. GOAL: A Generalist Combinatorial Optimization Agent Learner ICLR, 2025. paper

   Darko Drakulic, Sofia Michel, Jean-Marc Andreoli