Optimization of Weights in a Multiple Classifier Handwritten Word Recognition System Using a Genetic Algorithm

Abstract

Automatic handwritten text recognition by computer has a number of interesting applications. However, due to a great variety of individual writing styles, the problem is very difficult and far from being solved. Recently, a number of classifier creation methods, known as ensemble methods, have been proposed in the field of machine learning. They have shown improved recognition performance over single classifiers. For the combination of these classifiers many methods have been proposed in the literature. In this paper we describe a weighted voting scheme where the weights are obtained by a genetic algorithm.

Performance Based Modifications of Random Forest to Perform Automated Defect Detection for Fluorescent Penetrant Inspection

“…A genetic algorithm is used to actually determine an optimal combination of weight values. Comparing performance based weighting and genetic algorithms, it was found that the highest recognition rate was obtained with genetic weight optimisation [22].…”

“…A number of other modification methods have been investigated including modifications to feature selection [14,22,25], sampling methods [19,21,24], node splitting [15,20], performance evaluation [20], considering similarity between training examples [20,23,26] and others, details of which can be found in [15].…”

An automated supervised method for the diagnosis of Alzheimer’s disease based on fMRI data using weighted voting schemes

“…Another improvement of RF is accomplished by replacing majority voting with weighted voting. Since not all trees are equally responsible for incorrect classification of individual instances it is useful to use weights in the votes of the trees (RF with weighted voting) [22]- [27]. A description of all weighted voting schemes that were used is given in Section III.…”

“…Finally genetic algorithms (fifth weighted voting scheme) were used for the calculation of weights [27]. The weights of the chromosome with the highest fitness value, encountered during all generations, are the final result and are used for the weighted voting.…”

Weighted vote for trees aggregation in Random Forest

“…The second algorithm [20], the third [21] and the fifth algorithm [22] are based on the metrics between databases. The fourth [23] and the sixth [24] the use the weighting of trees according to their classification rate. The authors performed a number of experiments on the Al'zhamer databases to study the evolution of random forest performances when implementing the weighted vote or the classical majority vote.…”