“Bulletin Board”

 School of Mathematics - October 27, 2008

Mathematical Lecture

Somayeh Danafar
University of Tuebingen
Germany
November 5 & 6, 2008

 
 
Somayeh Danafar
University of Tuebingen
Germany
November 5 & 6, 2008



First talk:
Polar and Cartesian Models of Path Integration, Fitting a Quantitative Model to Experimental Data(November 5)



ABSTRACT: It is often assumed the navigation is used by animals indicates landmarks to locate the goal. However, many mammals including human and insects do homing by relying on self-motion cues. This process is known as Path integration. Path integration enables desert arthropods to find back to their nest on the shortest track from any position. We consider a mathematical model of a navigation system according to experiments conducted on humans in virtual environment in which only visual cues were provided. To perform path integration, forward moving speed and the angular turning rate are implemented into a linear system of differential equations in both Cartesian and Polar coordinate systems. Because in homing by triangle completion, homing distances were biased in the experimental results by subjects, we generate some Gaussian noises by Monte Carlo simulation for our mathematical model. The main objective of this work is fitting a quantitative model to the homing trajectory endpoints of experimental data, measure the goodness of fit, and find the best values of noise parameters. Here we used an extended Nelder-Mead simplex method as an optimization method to find the optimum values of noise parameters for our mathematical model.





Second talk:
Feed Forward Architecture Accounts for Rapid Categorization (after Tomas Serre, Andre Olive, and Tomaso Poggio)( November 6)

ABSTRACT: Primates are remarkably good at recognizing objects. The level of performance of their visual system and its robustness to image degradations still surpasses the best computer vision systems despite decades of engineering effort. In particular, the high accuracy of primates in ultra rapid object categorization and rapid serial visual presentation tasks is remarkable. Given the number of processing stages involved and typical neural latencies, such rapid visual processing is likely to be mostly feedforward. Here we show that a specific implementation of a class of feedforward theories of object recognition (that extend the Hubel and Wiesel simple-to complex cell hierarchy and account for many anatomical and physiological constraints) can predict the level and the pattern of performance achieved by humans on a rapid masked animal vs. non-animal categorization task.



Information
Time and Date: Wednsday, November 5, 2008 - 16:30-18:30
Thursday, November 6, 2008 - 10:30-12:00
Place: Lecture Hall, Niavaran Bldg., Niavaran Sqr., Tehran, Iran
    back to top
scroll left or right