Statistics Seminar: A Latent Factor Multivariate Point Process Model for Inferring the Neural Code

Presenter: Reza Ramezan, Associate Professor, Teaching Stream (Research Active), University of Waterloo

Description: Neural spike trains are electrochemical signals that are localized in time and serve as the primary means of communication among nerve cells. The analysis of simultaneously recorded spike trains from multiple neurons presents substantial statistical and computational challenges. In this talk, I will introduce a biologically inspired multivariate point process model for analyzing such data, formulated as a continuous-time latent factor model (LFM). To the best of our knowledge, this is the first continuous-time multivariate LFM developed to investigate neuronal interactions and functional connectivity using neural spike trains. By leveraging the computational advantages of approximate Bayesian inference, I will demonstrate that the proposed model can accommodate larger neuronal ensembles than existing methods, thereby providing an efficient framework for studying population coding. I will also discuss an application to experimental data from a classical conditioning study of the rat prefrontal cortex, illustrating how the proposed approach can enhance our understanding of the neural encoding of cue and outcome values.

About the speaker: Reza Ramezan is an Associate Professor, Teaching Stream (Research Active), in the Department of Statistics and Actuarial Science at the University of Waterloo, where he is also a core member of the institution’s Centre for Theoretical Neuroscience. His methodological research focuses on developing computationally efficient stochastic, multivariate, and high-dimensional models, with applications in neuroscience. Reza earned his PhD in Statistics from the University of Waterloo in 2014. He subsequently joined the Department of Mathematics at California State University, Fullerton, before returning to Waterloo in 2017. He has received numerous awards and recognitions for his teaching and research. His professional leadership includes serving as the former President of the Business and Industrial Statistics Section of the Statistical Society of Canada.

This seminar is hosted by Professor Yunyi Shen