University of Patras

About: University of Patras is a education organization based out in Pátrai, Greece. It is known for research contribution in the topics: Population & Catalysis. The organization has 13372 authors who have published 31263 publications receiving 677159 citations. The organization is also known as: Panepistímio Patrón.

Atomistic Simulation of Polymer Melt Elasticity: Calculation of the Free Energy of an Oriented Polymer Melt

Abstract: A method is developed for predicting the elasticity of a polymer melt through detailed atomistic simulations. The Helmholtz energy of a melt oriented by flow is postulated to be of the form A(T,ρ,c...

Random constraint satisfaction: a more accurate picture

Abstract: Recently there has been a great amount of interest in Random Constraint Satisfaction Problems, both from an experimental and a theoretical point of view. Rather intriguingly, experimental results with various models for generating random CSP instances suggest a "threshold-like" behavior and some theoretical work has been done in analyzing these models when the number of variables becomes large (asymptotic). In this paper we prove that the models commonly used for generating random CSP instances do not have an asymptotic threshold. In particular, we prove that as the number of variables becomes large, almost add instances they generate are trivially overconstrained. We then present a new model for random CSP and, in the spirit of random k-SAT, we derive lower and upper bounds for its parameters so that instances are "almost surely" underconstrained and overconstrained, respectively. Finally, for the case of one of the popular models in Artificial Intelligence we derive sharper estimates for the probability of being overconstrained, as a function of the number of variables. Canada PGS B Scholarship. E-mail: optas@cs.toronto.edu Partially supported by the EU ESPRIT Long-term Research Project ALCOM-IT (Project Nr. 20244). Supported in part by an NSERC grant. Partially supported by the EU ESPRIT Long-term Research Project ALCOM-IT (Project Nr. 20244).

Mixture of Expert Agents for Handling Imbalanced Data Sets

Abstract: Many real-world data sets exhibit skewed class distributions in which almost all cases are allotted to a class and far fewer cases to a smaller, usually more interesting class. A classifier induced from an imbalanced data set has, typically, a low error rate for the majority class and an unacceptable error rate for the minority class. This paper firstly provides a systematic study on the various methodologies that have tried to handle this problem. Finally, it presents an experimental study of these methodologies with a proposed mixture of expert agents and it concludes that such a framework can be a more effective solution to the problem. Our method seems to allow improved identification of difficult small classes in predictive analysis, while keeping the classification ability of the other classes in an acceptable level.