Temporal Graph Reproduction with RWIG

Sergey Shvydun; Anton David Almasan; Piet Van Mieghem

doi:10.1109/TNSE.2025.3555974

Temporal Graph Reproduction with RWIG

Sergey Shvydun^*, Anton David Almasan, Piet Van Mieghem

^*Corresponding author for this work

Network Architectures and Services

Research output: Contribution to journal › Article › Scientific › peer-review

3 Downloads (Pure)

Abstract

We examine the Random Walkers Induced temporal Graph (RWIG) model, which generates temporal graphs based on the co-location principle of M independent walkers that traverse the underlying Markov graph with different transition probabilities. Given the assumption that each random walker is in the steady state, we determine the steady-state vector s̃and the Markov transition matrix P _i of each walker w _i that can reproduce the observed temporal network G ₀, . . ., G _K _–1 with the lowest mean squared error. We also examine the performance of RWIG for periodic temporal graph sequences.

Original language	English
Pages (from-to)	3015-3024
Number of pages	10
Journal	IEEE Transactions on Network Science and Engineering
Volume	12
Issue number	4
DOIs	https://doi.org/10.1109/TNSE.2025.3555974
Publication status	Published - 2025

Bibliographical note

Green Open Access added to TU Delft Institutional Repository as part of the Taverne amendment. More information about this copyright law amendment can be found at https://www.openaccess.nl. Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Keywords

Generative Models
Markov Process
Network Dynamics
Random Walks
RWIG
Temporal Networks

Access to Document

10.1109/TNSE.2025.3555974

Temporal_Graph_Reproduction_With_RWIGFinal published version, 836 KB

Cite this

@article{e9b2fef5a44847a18a5dcaf5475bb475,

title = "Temporal Graph Reproduction with RWIG",

abstract = "We examine the Random Walkers Induced temporal Graph (RWIG) model, which generates temporal graphs based on the co-location principle of M independent walkers that traverse the underlying Markov graph with different transition probabilities. Given the assumption that each random walker is in the steady state, we determine the steady-state vector {\~s}and the Markov transition matrix P i of each walker w i that can reproduce the observed temporal network G 0, . . ., G K –1 with the lowest mean squared error. We also examine the performance of RWIG for periodic temporal graph sequences.",

keywords = "Generative Models, Markov Process, Network Dynamics, Random Walks, RWIG, Temporal Networks",

author = "Sergey Shvydun and Almasan, {Anton David} and Mieghem, {Piet Van}",

note = "Green Open Access added to TU Delft Institutional Repository as part of the Taverne amendment. More information about this copyright law amendment can be found at https://www.openaccess.nl. Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public. ",

year = "2025",

doi = "10.1109/TNSE.2025.3555974",

language = "English",

volume = "12",

pages = "3015--3024",

journal = "IEEE Transactions on Network Science and Engineering",

issn = "2327-4697",

publisher = "IEEE",

number = "4",

}

TY - JOUR

T1 - Temporal Graph Reproduction with RWIG

AU - Shvydun, Sergey

AU - Almasan, Anton David

AU - Mieghem, Piet Van

N1 - Green Open Access added to TU Delft Institutional Repository as part of the Taverne amendment. More information about this copyright law amendment can be found at https://www.openaccess.nl. Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

PY - 2025

Y1 - 2025

N2 - We examine the Random Walkers Induced temporal Graph (RWIG) model, which generates temporal graphs based on the co-location principle of M independent walkers that traverse the underlying Markov graph with different transition probabilities. Given the assumption that each random walker is in the steady state, we determine the steady-state vector s̃and the Markov transition matrix P i of each walker w i that can reproduce the observed temporal network G 0, . . ., G K –1 with the lowest mean squared error. We also examine the performance of RWIG for periodic temporal graph sequences.

AB - We examine the Random Walkers Induced temporal Graph (RWIG) model, which generates temporal graphs based on the co-location principle of M independent walkers that traverse the underlying Markov graph with different transition probabilities. Given the assumption that each random walker is in the steady state, we determine the steady-state vector s̃and the Markov transition matrix P i of each walker w i that can reproduce the observed temporal network G 0, . . ., G K –1 with the lowest mean squared error. We also examine the performance of RWIG for periodic temporal graph sequences.

KW - Generative Models

KW - Markov Process

KW - Network Dynamics

KW - Random Walks

KW - RWIG

KW - Temporal Networks

UR - http://www.scopus.com/inward/record.url?scp=105002025907&partnerID=8YFLogxK

U2 - 10.1109/TNSE.2025.3555974

DO - 10.1109/TNSE.2025.3555974

M3 - Article

AN - SCOPUS:105002025907

SN - 2327-4697

VL - 12

SP - 3015

EP - 3024

JO - IEEE Transactions on Network Science and Engineering

JF - IEEE Transactions on Network Science and Engineering

IS - 4

ER -

Temporal Graph Reproduction with RWIG

Abstract

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this