Applied math seminar

Event Type: 
Andrew Sornborger
Event Date: 
Monday, August 28, 2017 -
3:30pm to 4:30pm
General PublicFaculty/StaffStudents
Math and Stat Department

Event Description: 

Title: Pulse Gating and Neural Computing 

Abstract: High-fidelity information propagation can be crucial for information processing in neural systems. This is the case particularly in systems with no clocks, such as biological neuronal networks or analog neuromorphic systems. In this talk, I will discuss the synfire-gated synfire chain (SGSC), a pulse-gating mechanism that is capable of faithfully propagating graded (firing rate) information in networks of spiking neurons. After first presenting the SGSC and giving an understanding of why it is robust, I will show how SGSCs can be used to process information and make decisions, resulting in a framework that is sufficiently powerful for general computation. Then I will show how SGSCs can be deployed in concert with Hebbian synaptic plasticity in an algorithm capable of learning a stochastic process and making predictions of future process values.