Year: 2014
Type: dissertation
Abstract: This dissertation explores a novel statistical technique—information geometric method for theory and its application in analysis of multiple neuronal spike data. The previous studies have indicated that information-geometric method provides a powerful tool of estimating neuronal interactions from observed spiking data. However, these studies were conducted based on simplified neural network structure, which has limitations in the real brain. We systematically extended the previous studies by using intensive mathematical analysis and numerical simulations of realistic and complex neural network. The studies show that information geometric approach provide robust estimation for the sum of the connection weights between neuronal pairs in a complex recurrent network, providing a way of investigating the underlying network structures from neuronal spike data.
Author: Yimin Nie
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W2226424700