Network Structural Reconstruction Based on Delayed Transfer Entropy and Synthetic Data
Daniel Rodrigues de Lima1; Fernando Pasquini Santos1; Carlos Dias Maciel1
1 EESC - USP
Resumo
The knowledge of how signals are received, processed, and transmitted in neuronal systems is one of the bio-inspired engineering objectives. In this area, not only the physiology of separated neurons is relevant, but also, the connections among them, the neuronal topology. A modeling process of such biological system requires the integration of both, physiological and topological properties. However, there is a limitation in the modeling process due to the impossibility of recording all the system neurons at the same time. To solve this problem, we propose the usage of simulations and information theoretic measures to infer a network topology. Three test cases were simulated, and the interactions were measured with Transfer Entropy resulting in topology candidates. In two cases we could visually recover the connections from the graphs. In a third case we found a residual connection which allowed us to explore some properties from the network topology.
Palavras-chave: Bio-inspired engineering; neuronal systems; simulation; information theoretic measures; network topology