Adaptive Neuro Fuzzy Inference System, Neural Network and Support Vector Machine for Caller Behavior Classification

Abstract

A classification system that accurately categorizes caller behavior within Interactive Voice Response systems would assist in developing good automated self service applications. This paper details the implementation of such a classification system for a pay beneficiary application. Adaptive Neuro-Fuzzy Inference System (ANFIS), Feed forward Artificial Neural Network (ANN) and Support Vector Machine (SVM) classifiers were created. Exceptional results were achieved. The ANN classifiers are the preferred models. ANN classifiers achieved 100% classification on 'Say account', 'Say amount' and 'Select beneficiary' unseen data. The ANN classifier yielded 95.42% accuracy on 'Say confirmation' unseen data.