Learning to Configure Computer Networks with Neural Algorithmic Reasoning
Abstract
We present a new method for scaling automatic configuration of computer networks. The key idea is to relax the computationally hard search problem of finding a configuration that satisfies a given specification into an approximate objective amenable to learning-based techniques. Based on this idea, we train a neural algorithmic model which learns to generate configurations likely to (fully or partially) satisfy a given specification under existing routing protocols. By relaxing the rigid satisfaction guarantees, our approach (i) enables greater flexibility: it is protocol-agnostic, enables cross-protocol reasoning, and does not depend on hardcoded rules; and (ii) finds configurations for much larger computer networks than previously possible. Our learned synthesizer is up to 490x faster than state-of-the-art SMT-based methods, while producing configurations which on average satisfy more than 93% of the provided requirements.
Research Area: Verification and Synthesis
People
BibTex
@INPROCEEDINGS{beurer-kellner2022learning,
isbn = {978-1-7138-7108-8},
year = {2022},
booktitle = {Advances in Neural Information Processing Systems 35},
type = {Conference Paper},
editor = {Koyejo, Sanmi and Mohamed, Shakir and Agarwal, Alekh and Belgrave, Danielle and Cho, Kyunghyun and Oh, Alice},
institution = {ETHZ},
author = {Beurer-Kellner, Luca and Vechev, Martin and Vanbever, Laurent and Veličković, Petar},
abstract = {We present a new method for scaling automatic configuration of computer networks. The key idea is to relax the computationally hard search problem of finding a configuration that satisfies a given specification into an approximate objective amenable to learning-based techniques. Based on this idea, we train a neural algorithmic model which learns to generate configurations likely to (fully or partially) satisfy a given specification under existing routing protocols. By relaxing the rigid satisfaction guarantees, our approach (i) enables greater flexibility: it is protocol-agnostic, enables cross-protocol reasoning, and does not depend on hardcoded rules; and (ii) finds configurations for much larger computer networks than previously possible. Our learned synthesizer is up to 490x faster than state-of-the-art SMT-based methods, while producing configurations which on average satisfy more than 93% of the provided requirements.},
language = {en},
address = {Red Hook, NY},
publisher = {Curran},
title = {Learning to Configure Computer Networks with Neural Algorithmic Reasoning},
PAGES = {730 - 742},
Note = {36th Annual Conference on Neural Information Processing Systems (NeurIPS 2022); Conference Location: New Orleans, LA, USA; Conference Date: November 28 - December 9, 2022; Poster presentation on December 1, 2022.}
}Research Collection: 20.500.11850/589728