Search

In this paper, we propose a new class of neuro-fuzzy systems. Moreover, we develop a novel method for reduction of such systems without the deterioration of their accuracy. The reduction algorithm gradually eliminates inputs, rules, antecedents, and the number of discretization points of integrals in the center of area defuzzification method. It then automatically detects and merges similar input and output fuzzy sets. Through computer simulations it is shown that accuracy of the system after reduction and merging has not deteriorated despite the fact that in some cases up to 54% of various parameters and 74% of inputs were eliminated. The reduction algorithm has been tested using well-known classification benchmarks. [ABSTRACT FROM AUTHOR]

This paper presents original solutions from the field of intelligent expert systems for use in behavioral biometrics. They combine possibilities offered by biometric methods with the theory of fuzzy sets and the theory of population-based algorithms. Behavioral biometrics is concerned with learned behaviors such as a way of signing, movement, speaking, etc. However, these attributes trend to change over time. This is particularly important in a variety of fields including identity authentication. In this paper we present a new approach to the analysis of changes in behavioral characteristics. The purpose of the proposed approach is to predict values of features describing the so-called dynamic signature. It is the signature represented by waveforms describing a number of features including pen pressure and velocity. The proposed approach can be particularly useful in cases in which the interval between signature acquisition sessions is long. Our approach was tested with the use of the dynamic signature database called ATVS-SLT DB. It seems that the solutions proposed in this paper offer a new and interesting look at the solution of the problem of variability (aging) of biometric features. [ABSTRACT FROM AUTHOR]

Abstract: In the paper, the evolutionary strategy is applied for learning flexible neuro-fuzzy systems. In the process of evolution we determine: fuzzy inference (Mamdani type or logical type—described by an -implication), specific fuzzy implication, if the logical type system is found in the process of evolution or specific -norm connecting antecedents and consequences, if the Mamdani type system is found in the process of evolution, specific -norm for aggregation of antecedents in each rule, specific triangular norm describing aggregation operator, shapes and parameters of fuzzy membership functions, weights describing importance of antecedents of rules, and weights describing importance of rules, parameters of adjustable triangular norms, parameters of soft triangular norms. It should be noted that the crossover and mutation operators are chosen in a self-adaptive way. The method is tested using well known classification benchmarks. [Copyright &y& Elsevier]