CS700:Graduate Seminar in Computer Science & Informatics

A Supervised Learning algorithm, based on Lotka-Volterra models, with applications in Computational Intelligence in Finance

The talk is divided into two parts. The main part is on a supervised learning classification algorithm, based on an adaptation of the Generalized Lotka-Volterra model, which mathematically describes the evolution of the decision variables in the course of learning. Due to simplicity of the formulation, the model^?s convergence is given by a symmetric positive definite linear system. The model and associated algorithm feature implicit regulation of overfitting, non-linear classification through kernel functions, multi-class classification, low parameterization, and fast training. The second part is a demonstration of a software tool, called VolatilityAnalyst, that implements a methodology for detection and prediction of high volatility clusters in time series, using a supervised learning setup.
Karen Hovsepian is a Ph.D. candidate in the Computer Science department at New Mexico Tech. His interest includes Interaction Modeling Based Learning.