Sikun Lin successfully passed her MAE today on "Towards a Better Understanding of Brain Networks."
Committee: Ambuj Singh (Chair), Xifeng Yan, B.S. Manjunath, Tommy Sprague
Abstract:
Human brain is one of the most complex networks. It is anatomically organized over multiple spatial scales and is functionally interactive over multiple time scales. Understanding brain connectivity and activity, especially the relationship between the anatomical brain structures and the dynamics of neural processes, is a central question in neuroscience. An increasing number of theoretical and empirical studies approach the function of the human brain from a network perspective. This is made feasible by the development of new image acquisition techniques, large-scale initiatives using those techniques to collect population data, as well as new tools from graph theory, convex optimization, and deep learning (especially graph neural networks that take graphs as inputs). We need a better understanding of how brain structures give rise to various neuronal responses across multiple tasks as carried out by different individuals, as it is not only useful for neuroscience but also helpful in clinical practice and building more robust and human-like machines.
In this talk, I will introduce different brain data modalities, discuss optimization methods for mapping structural and functional brain networks, and cover graph neural networks that can be applied efficiently on brain data. I will also touch on how advances and challenges in neuroscience and artificial intelligence benefit each other, driving both fields forward.