Candidate: Wenhao Zhang, University of Pittsburg
Topic: Linking Theories of Computation with Neural Circuits in the Brain
The main goal of my research is bridging the abstract theories of computation
with concrete neural circuits in the brain. Studying this could provide theoretical
explanations of experimental findings, as well as provide new insight from neural
circuits to develop novel AI algorithms. In this talk, I’d like to present an overview
of my research paths in developing neural circuit models to implement Bayesian
and causal inference with a progression of network complexity. I will start from
my recent work of stimulus representation and inference in a recurrent circuit and
use it to give novel explanation on noise correlation (Cosyne 19, 20). Then I will
extend the single recurrent network to several coupled networks and use it to
achieve multisensory integration (Zhang, NIPS 13, JNS 16). At last, I will present
how I develop a circuit to implement the causal inference, i.e., integration vs.
segregation (Zhang, NIPS 16, 19; eLife 19), and my future plans along this path.
Join Zoom Here