Henri Jacques Suermondt

Using Genetic Programming to Obtain a Closed-Form Approximation to a Recursive Function

We demonstrate a fully automated method for obtaining a closedform approximation of a recursive function. This method resulted from a realworld problem in which we had a detector that monitors a time series and where we needed an indication of the total number of false positives expected over a fixed amount of time. The problem, because of the constraints on the available measurements on the detector, was formulated as a recursion, and conventional methods for solving the recursion failed to yield a closed form or a closed-form approximation. We demonstrate the use of genetic programming to rapidly obtain a high-accuracy approximation with minimal assumptions about the expected solution and without a need to specify problem-specific parameterizations. We analyze both the solution and the evolutionary process. This novel application shows a promising way of using genetic programming to solve recurrences in practical settings.

Probabilistic and Bayesian Representations of Uncertainty in Information Systems: A Pragmatic Introduction

A great deal has been written about the underlying principles of alternative methods of representing uncertainty—not the least about probability and Bayesian methods. While we cannot entirely resist discussing basic principles, we will focus on the pragmatic issues, which too often get lost under the mass of philosophy and mathematics. We will address such questions as: How can we use probability to represent the various types of uncertainty? How can we quantify these uncertainties? How much effort is necessary to do so? How can we obtain the greatest benefits from representing uncertainty while minimizing the effort? There are a variety of reasons to represent uncertainty and a variety of probabilistic and Bayesian ways to do so, requiring varying amounts of effort. We discuss an approach to resolve these issues, so that the costs will be commensurate with the benefits.