BIOMETRICS NEURAL NETWORKS METHODOLOGY

We demoralized notion of multi-modal biometrics, particularly ranked-level combination approach to design of extremely dependable and accurate biometric system. We also looked at compensation which can be gained by utilizing supplementary information about a subject, such as height, age and gender, or so-called soft biometric patterns. We also looked at the new research domain combining behavioral and appearance-based individuality one example of artificial entities, such as robots, avatars and intellectual software agents.

In this section, we look at learning approach and try to bring benefits of utilizing and identifying most prominent/significant patterns in multitude of biometric data, which not essential originates from the same biometric. Features from different biometric sources can be combined at either before-matching or after-matching stages and then most significant traits can be recognized through dimensional decrease or adaptive knowledge approaches.

To further understand the main benefits of this idea, it is important to understand the general architecture of the neural network based biometric recognition system. An illustration of the proposed multi-modal biometric system can be viewed in presents a flowchart for creating templates for training biometric system data set  For each of system N users, their individual biometrics are being collected and then represented as a reduced-dimensional feature vector set. This feature vector set will be then given as an input to neural network based on chaotic associative memory to learn common patterns for subsequent user recognition.

Artificial neural networks are computational models, where dispensation is performed by solid processing units. They were popularized in the works by Hop field (1990) and Choi and Hubernam (1983). Since then, research and applications of this paradigm became abundant for linear and nonlinear, static and dynamic systems (Wang & Shi, 2006; Yamada, Aihara, & Kotani, 1993; Yao, Freeman, Burke, & Yang, 1991). The nature of the network allows for a natural parallel computation and makes them an excellent learner as opposite to other traditional methodologies.

Artificial neural networks idea can be associated with the key work of McCulloch and Pitts (McCulloch & Pitts, 1943), which was published in 1943 in Bulletin of Mathematical Biophysics. This work is now considered to be a classical work in cognitive computational sciences domain. In 1987, Lippmann described how variety of applied troubles can be solved with using neural nets.

Chaos Online Dictionary definition is “complete disorder or confusion” or “Behavior so unpredictable as to appear random, owing to great sensitivity to small changes in conditions.” In artificial neural network research, one of the first scientists to take a close look at chaos in biological patterns.