PID5748855.pdf (239.78 kB)
Functional topology inference from network events
conference contribution
posted on 2023-06-09, 16:32 authored by Antoine Messager, George ParisisGeorge Parisis, Istvan Kiss, Robert Harper, Philip Tee, Luc BerthouzeLuc BerthouzeIn this paper we present a novel approach for inferring functional connectivity within a large-scale network from time series of emitted node events. We do so under the following constraints: (a) non-stationarity of the underlying connectivity, (b) sparsity of the time-series of events, and (c) absence of an explicit model describing how events propagate through the network. We develop an inference method whose output is an undirected weighted network, where the weight of an edge between two nodes denotes the probability of these nodes being functionally connected. Two nodes are assumed to be functionally connected if they show significantly more coincident or short-lagged events than randomly picked pairs of nodes with similar levels of activity. We develop a model of time-varying connectivity whose parameters are determined by maximising the model’s predictive power from one time window to the next. We assess the accuracy, efficiency and scalability of our method on a real dataset of network events spanning multiple months.
Funding
A fast method for calculating the proximity matrix in a large-scale dynamic network; G1742; MOOGSOFT INC; Agreement dated 17 December 2014
History
Publication status
- Published
File Version
- Accepted version
Journal
2019 IFIP/IEEE Symposium on Integrated Network and Service Management (IM)Publisher
Institute of Electrical and Electronics EngineersEvent name
IFIP/IEEE International Symposium on Integrated Network Management. Intelligent Management for the Next Wave of Cyber and Social NetworksEvent location
Washington DC, USAEvent type
conferenceEvent date
8-12 April 2019ISBN
9783903176157Department affiliated with
- Informatics Publications
Research groups affiliated with
- Sussex Neuroscience Publications
Full text available
- Yes
Peer reviewed?
- Yes