Improving Decision Effectiveness of Artificial Neural Networks: A Modified Genetic Algorithm Approach

Abstract

This study proposes the use of a modified genetic algorithm (MGA), a global search technique, as a training method to improve generalizability and to identify relevant inputs in a neural network (NN) model. Generalizability refers to the NN model's ability to perform well on exemplars (observations) that were not used during training (outof-sample); improved generalizability enhances NN's acceptability as a valid decisionsupport tool. The MGA improves generalizability by setting unnecessary weights (or connections) to zero and by eliminating these weights. Because the eliminated weights have no further impact on the training (in-sample or out-of-sample data), the relevant variables can be identified from the model. By eliminating unnecessary weights, the MGA is able to search and find a parsimonious model that generalizes well. Unlike the traditional NN, the MGA identifies the model variables that contribute to an outcome, helping decision makers to rationalize output and accept results with greater confidence. The study uses real-life data to demonstrate the use of MGA.

Bankruptcy Prediction Using Logit and Genetic Algorithm Models: A Comparative Analysis

“…Every genetic algorithm has a population of individuals; every individual indicates a possible solution to the problem ready. An individual can be demonstrated in many various ways regarding the problem to solve, the most popular way is, according to Sivanandam and Deepa (2008, p. 43), to indicate an individual as a bit-string Hekanaho (1997) and Sexton et al (2003) combined neural networks with a GA where the GA's role is as a feature selection process for the neural network. The GA model performed better than ANN, MDA, and logit, and the GA rules can be learned in parallel (Back et al 1996).…”

An Empirical Analysis of Financially Distressed Italian Companies

“…Despite the numerous empirical findings, different research concerns still need to be addressed, such as: the definition of failure, the stability of the data, the choice of the sample design, the variable selection (Amendola et al, 2011b;Hä rdle et al, 2009;Sensini, 2015;Sexton et al, 2003).…”

Predicting bankruptcy using neural networks and other classification methods: The influence of variable selection techniques on model accuracy

“…To overcome these limitations, some authors have explored other techniques such as genetic algorithms [11,26,27,12,28,13,29] or methods that fit a neural network [30,31,32,33]. But these examples are very few and no comparative study has analyzed the influence of a variable selection technique on the predictive performance of a model.…”

“…Sen et al [32] Tests applied to the weight of the neural network MLP-BP Serrano-Cinca [46] Variables used in one previous study MLP-BP Sexton et al [13] Genetic algorithm applied to variables commonly used in financial analysis MLP-GA Shin et Lee [71] Stepwise search with a criterion optimized for discriminant analysis and t test…”

“…Other authors have also analyzed the influence of sample size, group size, or the cost of misclassification [10]. Only a very few have attempted to assess the contribution of variable selection techniques to the performance of a model [11,12,13]. To date, to our knowledge, nobody has done a comparative analysis of variable selection methods.…”