Biometric Authentication: A Machine Learning Approach

7.2. Basic Decision-Based Neural Networks

In the binary classification problem, the pattern space is divided into two regions, and each class occupies its own region. In the clearly separable case, the two classes are separated by the decision boundary, defined as the hypersurface on which the two discriminant functions have equal scores. The objective of the learning phase is to determine the best discriminant functions, which in turn dictate decision boundaries.

A pioneering decision-based neural model is the perceptron originally proposed by Rosenblatt [322]. The (linear) perceptron was designed to separate two classes by a linear decision boundary (cf. Figure 7.1(a)). To deal with a more flexible decision boundary, the linear decision boundary ...