# A Deep Learning Perspective on Network Routing

arXiv: https://arxiv.org/abs/2303.00735 · HTML: https://mlsanity.com/paper/2303.00735/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 7 of its 60 references that are in the map.
- First posted on arXiv 1 Mar 2023.

## Builds on (most cited first, 7 of 7)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [A Comprehensive Survey on Graph Neural Networks](https://mlsanity.com/paper/1901.00596/) (2019)
- [Deep Reinforcement Learning that Matters](https://mlsanity.com/paper/1709.06560/) (2017)
- [Beyond Convexity: Stochastic Quasi-Convex Optimization](https://mlsanity.com/paper/1507.02030/) (2015)
- [Solving Large-Scale Granular Resource Allocation Problems Efficiently with POP](https://mlsanity.com/paper/2110.11927/) (2021)
- [Lying Your Way to Better Traffic Engineering](https://mlsanity.com/paper/1610.02728/) (2016)

## Built on by (most cited first, 1 of 1)

- [Pegasus: A Universal Framework for Scalable Deep Learning Inference on the Dataplane](https://mlsanity.com/paper/2506.05779/) (2025)

## Abstract (arXiv)

Routing is, arguably, the most fundamental task in computer networking, and the most extensively studied one. A key challenge for routing in real-world environments is the need to contend with uncertainty about future traffic demands. We present a new approach to routing under demand uncertainty: tackling this challenge as stochastic optimization, and employing deep learning to learn complex patterns in traffic demands. We show that our method provably converges to the global optimum in well-studied theoretical models of multicommodity flow. We exemplify the practical usefulness of our approach by zooming in on the real-world challenge of traffic engineering (TE) on wide-area networks (WANs). Our extensive empirical evaluation on real-world traffic and network topologies establishes that our approach's TE quality almost matches that of an (infeasible) omniscient oracle, outperforming previously proposed approaches, and also substantially lowers runtimes.