Online Graph Learning from Time-Varying Structural Equation Models

Online Graph Learning from Time-Varying Structural Equation Models

Natali, A.
Isufi, E.
Coutino, M.
Leus, G.

Matthews, M.B. (editor)

2021

Topology identification is an important problem across many disciplines, since it reveals pairwise interactions among entities and can be used to interpret graph data. In many scenarios, however, this (unknown) topology is time-varying, rendering the problem even harder. In this paper, we focus on a time-varying version of the structural equation modeling (SEM) framework, which is an umbrella of multivariate techniques widely adopted in econometrics, epidemiology and psychology. In particular, we view the linear SEM as a first order diffusion of a signal over a graph whose topology changes over time. Our goal is to learn such time-varying topology from stream ing data. To attain this goal, we propose a real-time algorithm, further accelerated by building on recent advances in time-varying optimization, which updates the time-varying solution as a new sample comes into the system. We augment the implementation steps with theoretical guarantees, and we show performances on synthetic and real datasets.

Dynamic topology identification
graph learning
graph signal processing
time-varying optimization
E-learning
Economics
Learning systems
Signal processing
Statistics
Time varying control systems
Time varying networks
Dynamic topologies
Dynamic topology identification
Graph learning
Graph signal processing
Optimisations
Signal-processing
Structural equation models
Time varying
Time-varying optimization
Topology identification
Topology

http://resolver.tudelft.nl/uuid:125c480a-d2b7-4309-a8c6-748782b00898

967245

IEEE Computer Society

9781665458283

1058-6393

Conference Record - Asilomar Conference on Signals, Systems and Computers, 55th Asilomar Conference on Signals, Systems and Computers, ACSSC 2021, 31 October 2021 through 3 November 2021, 1579-1585

conference paper