Research Fellows Directory
Mr Xiaozhao Zhao
I am a PhD student in computer science from Tianjin University, China. My present research has been focused on machine learning (ML) and information retrieval (IR). For ML, my recent research focuses on efficient model selection for neural networks, within the theoretical framework of Information Geometry. Specifically, a general parameter reduction criterion, called "confident information first principle", is proposed to automatically adapt neural networks to the given samples of data, so as to preserve the major information in data. For IR, I am interested in advanced methods to identify and extract the so-called "pure high-order word associations", also based on the Information Geometry, which have been shown effective to capture high-level semantics. I am also interested in the theoretical formalism of IR tasks using quantum theory, and have proposed novel document re-ranking and query expansion models for IR.