EVENT DETAILS AND ABSTRACT

Title: Efficient computation and cue integration with noisy population codes
Speaker: Alexandre Pouget
Speaker Info: University of Rochester
Brief Description:
Special Note: More current information may be available at Plan-it Purple
Abstract:

The brain represents sensory and motor variables through the activity of large populations of neurons. An important problem in neuroscience is to understand how the nervous system computes with these population codes, given that individual neurons are noisy and thus unreliable. I will focus on two general types of computation: function approximation and cue integration. I will show in particular that a class of neural networks, known as basis function networks with multidimensional attractors, can perform both types of computation optimally with noisy neurons. The architecture of these networks follows closely the one observed in the cortex, in that the networks use population codes and rely on feedforward, feedback and lateral connections to integrate information across layers. Moreover, neurons in the intermediate layers show response properties similar to those observed in several multimodal cortical areas. Thus, basis function networks with multidimensional attractors may be used by the brain to compute efficiently with population codes.