This is a past event.

Biomedical Engineering Research Seminar

Rodica Curtu, Ph.D.

Michigan Technological University

Abstract

Discovering dynamical patterns from high fidelity time series is typically a challenging task. In this talk I will discuss a data-driven method for the analysis of neural recordings taken from the auditory cortex of human subjects who listened to sequences of repeated triplets of tones and reported their perception by pressing a button. Subjects reported spontaneous alternations between two auditory perceptual states: a galloping-like rhythm (1-stream) and a Morse-code like rhythm (2-streams). Our algorithm leverages time-delayed coordinates, diffusion maps, and dynamic mode decomposition, to identify neural features in large scale brain recordings that correlate with subject-reported perception. The method captures the dynamics of perception at multiple timescales and distinguishes attributes of neural encoding of the stimulus from those encoding the perceptual states. Our analysis reveals a set of latent variables that exhibit
alternating dynamics along a low-dimensional manifold, like trajectories of attractor-based models.

Bio

I am by training an applied mathematician, but I have worked in the field of mathematical and computational neuroscience since graduate school. Soon after graduation, I took an academic appointment in Romania while, in parallel, I collaborated as an informal postdoc with scientists from the Center for Neural Science at New York University with expertise in computational neuroscience and the visual sciences. Since then, interdisciplinary collaborations have been a significant component of my research program with applications spanning cell biology, cognition, and visual and auditory perception. My lab has extensive experience in modeling neural data and in examining the dynamics of neural models. It applies nonlinear dynamical systems methods, pattern formation, computer simulations, statistical methods and machine learning algorithms to study a wide range of neural mechanisms underlying perception and cognition. My research group has also developed new methods for the analysis of large-scale data. These rely on diffusion maps and manifold learning principles to extract non-stationary features from multisite neuronal recordings.