"A Novel Optimal Configuration of Neural Networks by Multi-Objective Genetic Algorithm and Ensemble-classifier approach

Artificial Neural network is one of the most important data mining classification method among predictive algorithms. Performance of ANN affected by several parameters such as number of hidden layers neurons, learning function, stop conditions and network architecture. Parameter regulation is a point of critical challenge in this algorithm. The main purpose of this study is to provide a novel approach by using multi-objective genetic algorithm and ensemble classifier to obtain optimal parameters of ANN. First, a set of neural networks were trained by the settings their parameters through multi-objective genetic algorithm. Next, the best combination of neural networks was selected to make an ensemble classifier. This method was evaluated with five popular and available datasets. Three measurements; accuracy, time and ROC curve were considered to assess the efficiency. The experimental results show that the proposed approach can achieve a trade-off between time and accuracy by multi-objective genetic algorithm. Also using ensemble-classifiers approach increases the reliability of the model. Consequently, the proposed method enhances the accuracy and reduces the executing time.

کلیدواژه ها:

Artificial Neural network ، Multi-Objective Genetic Algorithm ، Ensemble-Classifier ، Optimal Parameters Regulation ،