← Back to Event List

Graduate Student Seminar

Location

University Center : 115

Date & Time

March 9, 2016, 11:00 am12:00 pm

Description

Session ChairJon Graf
DiscussantDr. Lo

Speaker 1: Zois Boukouvalis
Title
An Efficient Multivariate Generalized Gaussian Distribution Estimator: Application to IVA
Abstract
Due to its simple parametric form, multivariate Gaussian distribution (MGGD) has been widely used in modeling vector-valued signals.  Therefore, efficient estimation of its parameters is of significant interest for a number of applications. Independent vector analysis (IVA) is a generalization of independent component analysis (ICA) that makes full use of the statistical dependence across multiple datasets to achieve source separation, and can take both second and higher-order statistics into account. MGGD provides an effective model for IVA as well to model the latent multivariate variables--sources--and the performance of the IVA algorithm highly depends on the estimation of the source parameters. In this work, we propose an efficient estimation technique based on the implementation of successful Riemannian averages of the fixed point iterates and demonstrate its successful application to IVA. We quantify the performance of MGGD parameter estimation using RA-FP and further verify the effectiveness of the new IVA algorithm using simulations.

Speaker 2: Xinxuan Li
Title
A computational model of neurons in the thalamic reticular nucleus
Abstract
The thalamic reticular nucleus (TRN), a shell of GABAergic neurons that surrounds the dorsal thalamus, generates and sustains rhythmic oscillations called sleep spindles. Recent evidence has shown that TRN neurons are connected by chemical synapses and gap junctions. Moreover, the primary chemical synapses in TRN modulated by GABAa receptors, have recently been shown to be excitatory instead of inhibitory, as previously suggested. In order to understand how TRN neurons generate and sustain sleep spindles, we develop a Hodgkin-Huxley based computational model of a TRN neuron with the currents and parameters supported by recent experimental data. The resulting computational model of a single TRN neuron demonstrates bistable behavior. We also introduce the large T-window current model and the small T-window current model as a means of reflecting the known heterogeneity of TRN neurons. To understand the network properties of TRN, we connect two model neurons with excitatory GABAa , inhibitory  GABAb synapses, and gap junctions.  The computational model of excitatory GABAa synapses and gap junction are consistent with experimental results. We also explore computational models of GABAb synapses in TRN neuron. Our goal is to develop a computational model of a TRN neuron, consistent with known experimental results, that demonstrate known rhythmic behavior of both individual TRN neurons, and networks of TRN neurons.