Parts | Days | Selection | Search | Updates | Downloads | Help
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 2: Focus Session: Machine Learning for Complex Socio-economic Systems
SOE 2.4: Topical Talk
Monday, March 18, 2024, 10:30–10:50, MA 001
Prediction of processes on networks — •Piet Van Mieghem — Delft University of Technology, Delft, The Netherlands
I will talk about two related, but different problems in network science in ref 1 and ref 2 below. First (ref 1), given the nodal states of a process (e.g. a spreading process) on a fixed network over a time interval [0,T], can we predict that process at the time t > T ? Can we unravel the topology of the fixed network? Second (ref. 2), we consider a temporal network that has evolved over a time, defined by a sequence of consecutive graphs {G1,G2,…,GT}. We present a linear system identification method that is able to exactly emulate the sequence {G1,G2,…,GT}. Thus, our method reproduces the same outcomes of the process that determined the temporal graph at times in the past. Can our method predict the temporal graph at Gt with t > T?
References: 1) Prasse, B. and P. Van Mieghem, 2022, "Predicting network dynamics without requiring the knowledge of the interaction graph", Proceedings of the National Academy of Sciences (PNAS), Vol. 119, No. 44, e2205517119. (DOI:pnas.2205517119) 2) Shvydun, S. and P. Van Mieghem, 2023, "System Identification for Temporal Networks", IEEE Transactions on Network Science and Engineering, to appear. (DOI:10.1109/TNSE.2023.3333007)
Keywords: dynamic process on networks; prediction