Thinklab-SJTU/awesome-ml4co

What it solves

This project provides a curated, centralized collection of academic resources focused on using machine learning to solve combinatorial optimization (CO) problems. It helps researchers and practitioners navigate the rapidly growing field of applying neural networks and reinforcement learning to complex tasks like routing, scheduling, and graph analysis.

How it works

The repository acts as a comprehensive bibliography and knowledge base. It categorizes research papers by specific problem types (such as the Travelling Salesman Problem, Job Shop Scheduling, or Graph Matching) and includes a section for foundational survey papers. It also features a beta benchmarking tool that visualizes performance data from various research papers to help users compare different algorithmic approaches.

Who it’s for

This resource is designed for researchers, students, and practitioners interested in the intersection of machine learning and operations research. It is particularly useful for those looking to understand the current state-of-the-art in neural combinatorial solvers or seeking specific literature on applying AI to hard optimization challenges.

Highlights

• Extensive categorization of over 30 distinct combinatorial optimization problem types.
• A dedicated collection of foundational survey papers spanning over two decades of research.
• Beta-stage benchmark charts that visualize performance and runtime data across various research baselines.
• Clear identification of contributions from the SJTU-Thinklab research group.
• Links to open-source code implementations for many of the listed research papers.