Latent Representation of Neural Data (in person)
In this talk, I will discuss our ongoing project aimed at understanding the neural basis of memory. More specifically, I will focus on a unique electrophysiological experiment designed to address fundamental and unresolved questions about hippocampal function. The experiment involves using high-density electrophysiological techniques to record neural activity (spikes and local field potentials) in hippocampal region CA1 as rats perform an odor sequence memory task. To visualize and explore the underlying neural patterns, we have developed a set of latent representation learning methods based on deep learning techniques. Our findings suggest a fundamental function of the hippocampal network is to encode, preserve, and predict the sequential order of experiences. For a more formal statistical inference, we have then developed a new method based on latent factor Gaussian process models to quantify the underlying sequential structure of brain activity during the memory task. Our method provides a parsimonious representation of brain dynamic functional connectivity, which is believed to play an important role in many aspects of cognition. Our findings could lead to unprecedented insight into the neural mechanisms underlying memory.