A new hope for network model generalization

Alexander Dietmüller, Siddhant Ray, Romain Jacob, Laurent Vanbever
7/12/2022
cs.NIcs.LG

No official code found. Best implementation: Siddhant-Ray/Network-Traffic-Transformer (24 stars, MIT, last commit 2024-01-17).

Abstract

Generalizing machine learning (ML) models for network traffic dynamics tends to be considered a lost cause. Hence for every new task, we design new models and train them on model-specific datasets closely mimicking the deployment environments. Yet, an ML architecture called_Transformer_ has enabled previously unimaginable generalization in other domains. Nowadays, one can download a model pre-trained on massive datasets and only fine-tune it for a specific task and context with comparatively little time and data. These fine-tuned models are now state-of-the-art for many benchmarks. We believe this progress could translate to networking and propose a Network Traffic Transformer (NTT), a transformer adapted to learn network dynamics from packet traces. Our initial results are promising: NTT seems able to generalize to new prediction tasks and environments. This study suggests there is still hope for generalization, though it calls for a lot of future research.

View on arXivView PDF

Code Implementations(1)

Siddhant-Ray/Network-Traffic-TransformerCommunity100%
MIT

Network Traffic Transformer to learn network dynamics from packet traces. Learn fundamental dynamics with pre-training and fine-tune to multiple applications.

244C++, TeXMar 15, 20222 years agoMIT
network-simulatortransfer-learningtransformer

Similar Papers

QT-Routenet: Improved GNN generalization to larger 5G networks by fine-tuning predictions from queueing theory

85% match1 repo82022

Harnessing Generative Pre-Trained Transformer for Datacenter Packet Trace Generation

84% match2025

Towards a Transformer-Based Pre-trained Model for IoT Traffic Classification

84% match1 repo92024

netFound: Foundation Model for Network Security

83% match1 repo402023

Towards a graph-based foundation model for network traffic analysis

83% match2024

TrafficLLM: Enhancing Large Language Models for Network Traffic Analysis with Generic Traffic Representation

83% match1 repo4542025

Cite this paper

@article{dietmller2022hope,
  title  = {A new hope for network model generalization},
  author = {Alexander Dietmüller and Siddhant Ray and Romain Jacob and Laurent Vanbever},
  year   = {2022},
  eprint = {2207.05843},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2207.05843v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2207.05843.svg)](https://codeofpaper.com/paper/2207.05843)

Discussion