The emergent properties of information processing in Bow Tie complex networks

点击翻译按钮获取中文摘要

This project will work to develop novel measures and methods for identifying nodes in complex networks that will dictate the final (steady) states of the network, with a focus on (but not limited to) biological networks as a target network type. Complex networks are comprised primarily of two types of representation, discrete (Boolean) and continuous (Ordinary differential equation) networks. It has been noted that many biological systems display bow tie architectures. Such that there are three distinct layers in the network; an input component, a strongly connected core and an output component. Literature hypothesises that information processing is an emergent feature of bow tie networks. Currently methods utilise either dynamical or structural information to identify a set of control nodes of a given network. I will develop a novel method that will use both structural and dynamical measures via the implementation of a genetic algorithm to identify a set of nodes that will control a complex network. Particularly targeting the states of output nodes. Research will focus on the development of an algorithm decomposing complex networks utilizing dynamical features of a network such as maximum entropy and structural features i.e centrality, criticality or connectedness measures.A set of motifs (such as feedback loops, feed forward loops etc), will be established to further understand complex features of network structures. Motifs can be described as micro level effects with macro effects on networks. Particular attention will be paid to the strongly connected core of a network, as more complex motifs are likely to appear due to the higher edge density. Understanding the effects of specific motifs on the macro behaviour of the network may allow for new methods to dictate final states.In one sentence my research will develop novel measures and methods that utilise both structural and dynamical information to find nodes in complex networks that can dictate steady states. This will be achieved by developing understanding of the underlying motifs and using artificial intelligence techniques such as evolutionary algorithms to identify target nodes.