The prediction results of the final classification models (ensemble model and individual DNN classifier) of the proposed approach using the PSO algorithm and random method for the 5-fold cross-validation runs on the MNIST Dataset.

(Here, Avg and Std denote the average and standard deviation of the prediction results, Ensemble and Individual denote the prediction results achieved by the ensemble model and the individual DNN classifier, respectively, and Time1 and Time 2 denote the computational times of the training phases of the ensemble model and the individual DNN classifier, respectively.).