Saved in:
Bibliographic Details
Main Authors: Du, Chengze, Xu, Heng, Yu, Zhiwei, Liu, Bo, Li, Jialong
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.15251
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912718613643264
author Du, Chengze
Xu, Heng
Yu, Zhiwei
Liu, Bo
Li, Jialong
author_facet Du, Chengze
Xu, Heng
Yu, Zhiwei
Liu, Bo
Li, Jialong
contents Network tomography aims to infer hidden network states, such as link performance, traffic load, and topology, from external observations. Most existing methods solve these problems separately and depend on limited task-specific signals, which limits generalization and interpretability. We present PLATONT, a unified framework that models different network indicators (e.g., delay, loss, bandwidth) as projections of a shared latent network state. Guided by the Platonic Representation Hypothesis, PLATONT learns this latent state through multimodal alignment and contrastive learning. By training multiple tomography tasks within a shared latent space, it builds compact and structured representations that improve cross-task generalization. Experiments on synthetic and real-world datasets show that PLATONT consistently outperforms existing methods in link estimation, topology inference, and traffic prediction, achieving higher accuracy and stronger robustness under varying network conditions.
format Preprint
id arxiv_https___arxiv_org_abs_2511_15251
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PLATONT: Learning a Platonic Representation for Unified Network Tomography
Du, Chengze
Xu, Heng
Yu, Zhiwei
Liu, Bo
Li, Jialong
Machine Learning
Networking and Internet Architecture
Network tomography aims to infer hidden network states, such as link performance, traffic load, and topology, from external observations. Most existing methods solve these problems separately and depend on limited task-specific signals, which limits generalization and interpretability. We present PLATONT, a unified framework that models different network indicators (e.g., delay, loss, bandwidth) as projections of a shared latent network state. Guided by the Platonic Representation Hypothesis, PLATONT learns this latent state through multimodal alignment and contrastive learning. By training multiple tomography tasks within a shared latent space, it builds compact and structured representations that improve cross-task generalization. Experiments on synthetic and real-world datasets show that PLATONT consistently outperforms existing methods in link estimation, topology inference, and traffic prediction, achieving higher accuracy and stronger robustness under varying network conditions.
title PLATONT: Learning a Platonic Representation for Unified Network Tomography
topic Machine Learning
Networking and Internet Architecture
url https://arxiv.org/abs/2511.15251