Bipin Kumar Tripathi

Biometric recognition by hybridization of evolutionary fuzzy clustering with functional neural networks

Computational intelligence is an emerging area having caliber to solve many real world complex problems. Proper synergism of evolutionary, fuzzy and neural techniques can be more suitable for solving these problems. A novel approach for human recognition which is based on the fusion of evolutionary fuzzy clustering and functional modular neural networks (FMNN) is presented in this paper. Here, evolutionary fuzzy clustering with Minkowski distance (EFC-MD) is proposed for pre-classification task which allocates training patterns into optimal number of clusters. Considering Minkowski distance matrices instead of Euclidian distance provide flexibility to clustering algorithm in acquiring any shapes for clusters. The functional modular neural network is trained according to the fuzzy distribution of patterns in a cluster by EFC-MD. The functional neural network discriminates itself with the conventional neural network in the context it process and classifies the patterns on the basis of fuzzy distribution of training patterns. Final recognition or identification of patterns is based on combined outcomes of FMNN. The experimental results present the efficacy of proposed technique and compare it with the recent research outcomes in related areas. The motivation of this fusion is demonstrated through four benchmark biometric datasets.

Evolutionary fuzzy clustering and functional modular neural network-based human recognition

Computational intelligence shows its ability for solving many real-world problems efficiently. Synergism of fuzzy logic, evolutionary computation, and neural network can lead to development of a computational efficient and performance-rich system. In this paper, we propose a new approach for solving the human recognition problem that is the fusion of evolutionary fuzzy clustering and functional modular neural networks (FMNN). Evolutionary searching technique is applied for finding the optimal number of clusters that are generated through fuzzy clustering. The functional modular neural network has been used for recognition process that is evaluated with the help of integration based on combining the outcomes of FMNN. Performance of the proposed technique has been empirically evaluated and analyzed with the help of different parameters.

An evolutionary fuzzy clustering with minkowski distances

In this paper, we present a novel evolutionary fuzzy clustering approach with Minkowski distances. Fuzzy clustering plays an important role for various kinds of classification problems. Evolutionary algorithm is used for searching the best partitioning among the populations generated by different runs of the fuzzy clustering algorithm. Evolutionary fuzzy clustering performs better as compared to the conventional fuzzy clustering in terms of classification accuracy and partitioning. Fuzzy c-means (FCM) is a data clustering algorithm in which each data point is associated with a cluster through a membership degree. Here, Minkowski distance is used with FCM instead of conventional Euclidian distance because of its more generalized nature. It does not restrict the shape of the clusters generated. Empirical evaluation demonstrates the performance of proposed novel technique in terms of precision and accuracy in various benchmark problems.

Human recognition based on oculo-motion characteristics

This paper is devoted largely for building a novel approach for human recognition using eye movement analysis. Velocity and dispersion threshold based fixation identification algorithms are employed for processing the raw scan path signals in oculo-motion matrices. A new hybrid intelligent model is deployed for classification over data retrieved from scan-path signals. Experimental results demonstrate the endeavor of oculo-motion signals as an effective biometric trait. This paper also demonstrates the relative comparison of the two fixation identification techniques combined with hybrid intelligent model.

Functional mapping with complex higher order compensatory neuron model

The basic ideas to develop artificial neural network (ANN) were originated with the investigation of brain0s micro-structure. It has been a steady endeavor in the research that followed to develop it further and integrate additional discoveries about the human brain with a view to evolve the artificial neuron model closer to the actual brain functioning. The pursuit has ever been on to replicate the typical characteristic of the neuron. The neuron response to the input signals impinged onto it, is defined how they are aggregated with in the unit. A substantial body of evidence has grown to support the presence of non-linear integration of synaptic inputs in the neuron cells. Superior functionality of ANN in complex domain has been observed in recent researches, which presented the second generation of development in ANN. In this paper, we explore the functional capabilities of a compensatory neuron model with complex-valued high order non-linear aggregation function. The strength and effectiveness of considered neuron is evaluated with an efficient learning algorithm in a complex domain. The performance analysis is carried out through a solid set of simulations.

A HYBRID INTELLIGENT MODEL BASED ON EVOLUTIONARY FUZZY CLUSTERING AND SYNDICATE NEURAL NETWORKS

In this article, a new hybrid intelligent model comprising a cluster allocation and adaptation component is developed for solving classification and pattern recognition problems. Its computation ability has been verified through various benchmark problems and biometric applications. The proposed model consists of two components: cluster distribution and adaptation. In the first module, mean patterns are distributed into the number of clusters based on the evolutionary fuzzy clustering, which is the basis for network structure selection in next module. In the second module, training and subsequent generalization is performed by the syndicate neural networks (SNN). The number of SNNs required in the second module will be same as the number of clusters. Whereas each network contains as many output neurons as the maximum number of members assigned to each cluster. The proposed novel fusion of evolutionary fuzzy clustering with a neural network yields superior performance in classification and pattern recognition problems. Performance evaluation has been carried out over a wide spectrum of benchmark problems and real-life biometric recognition problems with noise and occlusion. Experimental results demonstrate the efficacy of the methodology over existing ones.

Root-Power Mean Aggregation-Based Neuron in Quaternionic Domain

ABSTRACT This paper illustrates the new structure of artificial neuron based on root-power mean (RPM) aggregation for quaternionic-valued signals and also presented an efficient learning process of neural networks with quaternionic-valued RPM neurons. The main aim of this neuron is to present the potential capability of a nonlinear aggregation operation on the quaternionic-valued signals in neuron cell. A wide spectrum of aggregation ability of RPM in between minima and maxima has a beautiful property of changing its degree of compensation in the natural way which emulates the various existing neuron models as its special cases. Further, the quaternionic resilient propagation algorithm (ℍ-RPROP) with error-dependent weight backtracking step significantly accelerates the training speed and exhibits better approximation accuracy. The wide spectra of benchmark problem are considered to evaluate the performance of proposed quaternionic RPM neuron with ℍ-RPROP learning algorithm.

Higher-Order Computational Model for Novel Neurons

Artificial neural network (ANN) has attracted a tremendous amount of interest for the solution of many complicated engineering and real-life problems. A small complexity, quick convergence, and robust performance are vital for its extensive applications. These features are pertinent upon the architecture of the basic working unit or neuron model, used in neural network. The computational capability of a neuron governs the architectural complexity of its neural network, which in turn defines the number of nodes and connections. Therefore, it is imperative to look for some neuron models, which yield ANN having small complexity in terms of network topology, number of learning parameters (connection weights) and at the same time they should possess fast learning, and superior functional capabilities. The conventional artificial neurons compute its internal state as the sum of contributions (aggregation) from impinging signals. For a neuron to respond strongly toward correlation among inputs, one must include higher-order relation among a set of inputs in their aggregation. A wide survey into design of artificial neurons brings out the fact that a higher-order neuron may generate an ANN which can have better classification and functional mapping capabilities with comparatively less number of neurons. Adequate functionality of ANN in a complex domain has also been observed in recent researches. This chapter presents higher-order computational models for novel neurons with well-defined learning procedures. Their implementation in a complex domain will provide a powerful scheme for learning input/output mapping in complex as well as in real domain along with better accuracy in wide spectrum of applications. The real domain implementation may be realized as its special case. The purpose of investigation in this chapter is to present the suitability and sustainability of higher-order neurons for readers, which can serve as a basis of the formulation for powerful ANN.

Hybrid Computation Model for Intelligent System Design by Synergism of Modified EFC with Neural Network

In recent past, it has been seen in many applications that synergism of computational intelligence techniques outperforms over an individual technique. This paper proposes a new hybrid computation model which is a novel synergism of modified evolutionary fuzzy clustering with associated neural networks. It consists of two modules: fuzzy distribution and neural classifier. In first module, mean patterns are distributed into the number of clusters based on the modified evolutionary fuzzy clustering, which leads the basis for network structure selection and learning in associated neural classifier. In second module, training and subsequent generalization is performed by the associated neural networks. The number of associated networks required in the second module will be same as the number of clusters generated in the first module. Whereas, each network contains as many output neurons as the maximum number of members assigned to each cluster. The proposed hybrid model is evaluated over wide spectrum of benchmark problems and real life biometric recognition problems even in presence of real environmental constraints such as noise and occlusion. The results indicate the efficacy of proposed method over related techniques and endeavor promising outcomes for biometric applications with noise and occlusion.

Periocular Biometric Recognition Using Supervised Fuzzy Clustering

Abtsract—Developing a potential biometrics has been a key focus of research in recent years. Periocular biometrics is a new trait to deal with non-ideal scenarios in face and iris biometrics. It can be used as an alternative to iris recognition, if the iris images are captured at a distance. In forensic applications, this trait can be used individually as well as with other traits (face and iris) for effective and accurate identification. In recent researches, the periocular biometrics is significantly impacting the iris and face based recognition. In this paper, we investigated the efficacy of supervised fuzzy clustering for strict periocular region which does not involve the eyebrows. The fixed initialization is considered in proposed supervised fuzzy clustering instead of random initialization. Then fuzzy clustering motivated with partition index maximization is used to optimize the objective function, hence yield clusters with representative prototype. The fuzzy clustering is further generalized with Minkowski distance matrices to yield variable cluster shape. Recognition is done based on the minimum distance measure between the test patterns and the centroid of the clusters. We use eight hundred periocular region images extracted from AR face dataset of 40 subjects. Performance of the proposed technique has been evaluated in terms of rank-one and rank-two recognition accuracy. Experimental analysis demonstrates the efficacy of presented technique over other variants of fuzzy clustering techniques.