Regularization of a neural network for binary classification

Abstract

Neural networks are complex non-linear models used to learn representations of data with multiple levels of abstraction. In this work, we introduce the basic notions of a neural network model, explaining the training process and generalization capabilities. In particular, to avoid overfitting to the training data, we study the L2 regularization method. Finally, an application of 2D vector classification has been developed to illustrate the concepts