A Hybrid Neuro-Wavelet Based Pre-Processing Technique for Data Representation
Tej Singh, Dinesh Kumar Vishwakarma · 2017
This work is an effort to assess the learning and generalization performance of multilayer perceptron preceded by a preprocessing stage. Back-propagation algorithm is the best-known algorithm for supervised learning. This method used the gradient descent to calculate the loss function concerning all weight in the network. The performance function is calculated regarding mean square error (MSE). This technique is used to train for two different real-time data called Iris data and User Student Modeling data respectively. This BP Algorithm is trained for various combination of learning rate or (0.1 to1), and constant momentum mc (0.1-1.0). The performance function is calculated regarding mean square error (MSE). Principal Component Analysis (PCA) and Transfer cluster analysis (TCA) is applied to same data sets. The modifying data set is again trained with BP algorithm, and results are shown again by Box-Plot. Next, a continuous wavelets transform technique is after applying PCA and TCA technique on raw dataset. The obtained dataset is trained with BP algorithm for various types of combination of learning rates and momentum constants. The results showed that data preprocessing has a profound effect on classification performance of the final classifier.