Biometric Authentication: A Machine Learning Approach

7.3. Hierarchical Design of Decision-Based Learning Models

The basic DBNN adopts a simple OCON structure, and K subnets are used for the classification of K categories. To facilitate a more flexible learning machine model, a two-level hierarchical network is constructed by embedding GMM into the OCON framework. More exactly, the top level still retains the basic OCON structure as shown in Figure 7.3, while in the second level each subnet is further modeled by a GMM structure, resulting in a more sophisticated hierarchical structure as shown in Figure 7.5.

Figure 7.5. The hierarchical structure of decision-based neural networks. The basic OCON structure is adopted in the top level of the hierarchy. In the next sublevel, each subnet is further ...