Talk title: Using Neural Networks to Explore Waddington's Landscape

Speaker: Madhav Mani, PhD
Associate Professor of Engineering Sciences and Applied Mathematics
Northwestern University

Abstract: 

Inspired by the mysteries and structural connectivity of the brain, John Hopfield developed a general class of learning networks, which are both used as a model to understand brain function and mimic its computational capabilities for an enormous diversity of uses. Here, I will use as inspiration Waddington's metaphor for the robustness of evolved systems behaving as a ball rolling down in a canalized landscape that is held up by a complex network of genetic factors. I will present 3 recent investigations from our team (1, 2, and 3) where we leverage neural networks to study the origins and consequences of this powerful dynamical systems analogy in the context of the dynamics of cellular, embryonic, and MNIST digit differentiation.

Speaker Bio:

Madhav was born in India, and grew up across India, England, and now America. After an undergraduate degree in mathematics and physics at Cambridge University he completed a PhD with Michael Brenner and L. Mahadevan at Harvard University in soft matter physics. He switched to studying living systems during his postdoctoral work as a Simons Fellow at the Kavli Inst. of Theoretical Physics under the guidance of Boris Shraiman. Since then, he has been in applied mathematics at Northwestern, working with a team of fantastic graduate students and postdocs on a diversity of topics including Cell and Developmental biology, and ecology.

Host:
Milo Lin, PhD