GES-TSP: Graph Edge Sparsification for TSP
Quick Answer
The GES method introduces a learning-based graph edge sparsification approach for the Traveling Salesman Problem (TSP), achieving up to 95% edge pruning on the MATILDA dataset while maintaining a solution gap of less than 1%.
Quick Take
The GES method introduces a learning-based graph edge sparsification approach for the Traveling Salesman Problem (TSP), achieving up to 95% edge pruning on the MATILDA dataset while maintaining a solution gap of less than 1%. This adaptive method leverages geometric structural information to enhance computational efficiency, demonstrating over 99% pruning on large-scale TSPLIB instances with similar optimality results.
Key Points
- GES achieves up to 95% edge pruning on the MATILDA dataset.
- The method maintains a solution gap within 1% of the optimal value.
- It utilizes geometric structural information for adaptive sparsification.
- On TSPLIB instances, pruning rates exceed 99% with optimality gap below 1%.
- This approach significantly accelerates the solving process for TSP.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Solving large-scale instances of the Traveling Salesman Problem (TSP) exactly is computationally expensive. Researchers often employ graph sparsification methods to improve computational efficiency. Traditional sparsification methods typically rely on fixed heuristics and fail to fully exploit instance-specific structural information. In this paper, we propose Graph Edge Sparsification (GES), a learning-based sparsification approach for Euclidean TSP. By incorporating geometric structural information and combinatorial optimization technology, our proposed method adaptively generates a sparsification graph for different instances, significantly reducing the graph size and accelerating the solving process. Experimental results demonstrate that our sparsification method can prune up to 95% of edges on the MATILDA dataset, while keeping the solution gap within 1% of the optimal value. Moreover, our approach exhibits strong generalization capability on the TSPLIB this http URL some large-scale instances, the pruning rate exceeds 99%, while the optimality gap remains below 1%.
| Subjects: | Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Combinatorics (math.CO) |
| Cite as: | arXiv:2607.09708 [cs.AI] |
| (or arXiv:2607.09708v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.09708 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Tianfeng Chen [view email]
[v1]
Tue, 23 Jun 2026 11:13:29 UTC (2,779 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →Adversarial Social Epistemology for Assemblies of Humans and Large Language Models
The paper introduces Adversarial Social Epistemology (ASE) to analyze how agents manipulate trust in public communications, highlighting mechanisms that undermine the reliability of testimony and inference. It critiques existing frameworks like epistemic bubbles and misinformation diffusion, proposing a new language for understanding trust breaches and auditing inferential chains in densely interactive environments involving humans and large language models.