ABSTRACT
Pattern recognition involves a set of processes to define similarities and/or differences between two or more patterns. Patterns or data must be evaluated or measured to find distinctive characteristics. The first step in any pattern recognition scheme is to identify measurable quantities or characteristics of patterns that match a specific class of data. These measurable quantities are known as features. According to Theodoridis and Koutroumbas [85], features can be defined as a set of measurements used for recognition and classification. These measurements form a feature vector that is used for recognition purposes. In image recognition, examples of features include colors, edges, and spectrum frequencies. Pattern recognition, as described in the previous chapters, is a series of
processes including data acquisition, data pre-processing, and classification [86]. Each data presented for recognition is assigned to the data class that most closely matches the features of the data. These features are extracted before any classification/recognition process takes place. The extraction process is performed during the pre-processing stage of pattern recognition. In existing pattern recognition schemes, the number of features used tends to be very large. A phenomenon known as the “curse of dimensionality” arises as a result of the high dimensionality of the computational space. This chapter focuses on pattern recognition schemes involving multiple fea-
tures. A multiple-feature implementation enables a holistic approach to the pattern recognition procedure that takes into consideration all significant features representing a particular set of patterns, such as images and sensor readings. This multi-feature consideration is important when considering complex data in an Internet-scale environment. The multi-feature approach was designed to reduce the bias effect caused by selecting only a single feature for classification/recognition purposes. To avoid the curse of dimensionality, current approaches in pattern recognition require a significant amount of effort to analyze different forms of features. This effort limits their ability to seamlessly and effectively perform recognition and classification on complex data sets. Furthermore, the computational complexity of most existing schemes inhibits their ability to scale up to an increasing number of features. It is envisioned that the distributed approach can be implemented in
Internet-scale pattern recognition involving multiple features. It is argued that
a set of distributed computational networks working collaboratively can scale the pattern recognition scheme in response to an increasing number of features. In addition, the performance of this multi-feature scheme can be improved by a single-cycle learning distributed pattern recognition algorithm, such as the DHGN. In contrast to other contemporary machine learning approaches, our approach allows induction of new patterns in a fixed number of steps. While doing so, it exhibits a high level of scalability, i.e., the performance and accuracy do not degrade as the number of stored patterns increases. The pattern recognition capability remains comparable with contemporary approaches, such as the support vector machine (SVM), self-organizing map (SOM), and artificial neural network (ANN). Furthermore, all computations are completed within the pre-defined number of steps. The one-shot learning in this method is achieved by sidestepping the commonly used error/energy minimization and random walk approaches. The network functions as a matrix that holds all possible solutions for the problem domain. The DHGN approach finds and refines the initial solution by passing the results through a pyramidal hierarchy of similar arrays. In doing so, it eliminates/resolves pattern defects; distortions up to 20% are tolerated [64]. Previously encountered patterns are revealed and new patterns are memorized without the loss of stored information. In fact, the pattern recognition accuracy continues to improve as the network processes more sensory inputs [3].To achieve this goal, the DHGN distributed pattern recognition algorithm is extended for multifeature recognition and the analysis of complex data.
