The conventional real-valued neuro-fuzzy method (RNF) is based on classic fuzzy systems with antecedent membership functions and consequent singletons. Rules in RNF are made by all the combinations of membership functions; thus, the number of rules as well as total parameters increase rapidly with the number of inputs. Although network parameters are relatively less in the recently developed complex-valued neuro-fuzzy (CVNF) and quaternion neuro-fuzzy (QNF), parameters increase with number of inputs. This study investigates simplified fuzzy rules that constrain rapid increment of rules with inputs; and proposed simplified RNF (SRNF), simplified CVNF (SCVNF) and simplified QNF (SQNF) employing the proposed simplified fuzzy rules in conventional methods. The proposed simplified neuro-fuzzy learning methods differ from the conventional methods in their fuzzy rule structures. The methods tune fuzzy rules based on the gradient descent method. The number of rules in these methods are equal to the number of divisions of input space; and hence they require significantly less number of parameters to be tuned. The proposed methods are tested on function approximations and classification problems. They exhibit much less execution time than the conventional counterparts with equivalent accuracy. Due to less number of parameters, the proposed methods can be utilized for the problems (e.g., real-time control of large systems) where the conventional methods are difficult to apply due to time constrain.

The paper presents the method of medical images similarity estimation based on feature extraction and analysis. The proposed method has been developed for and tested on rat brain histological images, however, it can be applied for other types of medical images, since the general approach is based on consideration of the shape of core components present in a given template image. The proposed method can be used in image analysis tools in a wide range of image-based medical investigations, in particular, in the brain researches.
The theoretical background of the proposed method is presented in the paper. The expert evaluation approach used for assessment of the proposed method effectiveness is explained and illustrated by examples. The method of medical images similarity estimation based on feature analysis consists of several stages: colour model conversion, image normalization, anti-noise filtering, contours search, conversion, and feature analysis. The results of the proposed method algorithmic realization are demonstrated and discussed.

This work aims at designing a fractional Proportional-Integral-Derivative controller wherein we hybridize a genetic algorithm based fractional Proportional-Integral-Derivative controller with a fuzzy logic Proportional-Integral-Derivative controller. We attempt at optimizing the fractional order Proportional-Integral-Derivative controller parameters by incorporating a Genetic Algorithm based mechanism. Thereafter, the optimized genetic algorithm based fractional Proportional-Integral-Derivative control is further fine tuned and hybridized to a fuzzy Proportional-Integral-Derivative control. Here, fuzzy logic based inference mechanism is used to tackle system uncertainties and use of rule firing strengths produces an adaptive control. Genetic Algorithm has been used to generate the most optimal controller by a natural selection of the fittest. Amalgamating Genetic Algorithm and fuzzy control approaches on fractional order systems produces a highly efficient and noise tolerant control regime. We give simulation results and compare our hybrid approach against conventional and fractional Proportional-Integral-Derivative approaches on various integer and fractional order systems (with dead time) to elucidate its superiority and effectiveness.

Artificial neural networks (ANN) have been widely used in classification. They are complicated networks due to the training algorithm used to fix their weights. To achieve better neural network performance, many evolutionary and meta-heuristic algorithms are used to optimize the network weights. The aim of this paper is to implement recently evolutionary algorithms for optimizing neural weights such as Grass Root Optimization (GRO), Artificial Bee Colony (ABC), Cuckoo Search Optimization (CSA) and Practical Swarm Optimization (PSO). This ANN was examined to classify three classes of EEG signals healthy subjects, subjects with interictal epilepsy seizure, and subjects with ictal epilepsy seizures. The above training algorithms are compared according to classification rate, training and testing mean square error, average time, and maximum iteration.

Textural feature extraction of image is a basic work for image analysis. A number of approaches have been put forward to describe texture features quantitatively, such as gray level co-occurrence matrix, fractal wavelet, Gabor wavelet and local binary pattern etc, among them texture feature extracted based on “tuned” mask will not suffer from rotation and scale of images. However, it needs to take a lot of time to learn the tuned mask with the traditional methods and could not acquire the satisfying mask sometimes. In essence, it is a very hard combinational optimization problem and easy to fall into the local optimum with mountain climbing method. Bat algorithm is a newly proposed meta-heuristic optimization, which is employed to tune the optimal mask in the paper. The procedure of bat algorithm to learn the tuned mask is detailed. Experiments results testifies that the proposed method is propitious to draw texture features, its performance is better than the simple particle swarm optimization and genetic algorithm based mask tuning scheme, which is a robust approach for texture image analysis.

The objective of this paper is to improve the passenger ride comfort and safety in an active quarter car model. For this purpose, a quarter car model with passenger body and seat is considered to capture the dynamic behaviour of a real complete car system. To achieve the desired target, two different controllers such as Adaptive Neuro Fuzzy (ANFIS) controller and Hybrid ANFIS PID controller (HANFISPID) are designed. The controllers selection and design was aimed to achieve good passenger ride comfort and health, taking passenger body acceleration and displacement response under random road excitations. The performance of designed controllers are evaluated using simulation work in time and frequency domain. Simulation results show that the proposed HANFISPID control scheme can succesfully achieve the desired ride comfort and safety of passenger compared to passive and ANFIS controlled cases in an active quarter car model.

Teaching-Learning-Based Optimization (TLBO) is a contemporary algorithm being used as a novel, trustworthy, precise and robust optimization technique for global optimization over continuous spaces both constrained and unconstrained tribulations. TLBO works on the beliefs of teaching and learning and clearly justifies this pedagogy by highlighting the effect of power of a teacher on the output of learners in a class. This paper, explores the applicability of k-means unsupervised learning into TLBO with two endeavors, i.e. to automatically find the optimal number of naturally classified partition in the data without any prior information, and the other is to inspect the naturally classified partitions with cluster validity indices (CVIs) and endorse the goodness of clusters. The proposed automatic clustering algorithm using TLBO (AutoTLBO) pursues a novel evolutionary approach by incorporating the simple k-means algorithm and CVIs into TLBO to configure and validate automatic natural partition in datasets. This algorithm retains the core ideology of clustering to minimize the inter cluster distances and maximize the intra cluster distances among the data. Experimental analysis substantiates the openness of the anticipated method after inspecting suavest panoramic rendering over artificial and benchmark datasets.

Customer retention is becoming a key success factor for many business applications due to increasing market competition. Especially telecom companies are facing this challenge with a rapidly increasing number of service providers. Hence there is need to focus on customer churn prediction in order to detect the customers that are likely to churn i.e. switch from one service provider to another. Several data mining techniques are applied for classifying customers into the churn and non-churn category. But churn prediction applications comprise an imbalanced distribution of the dataset.
One of the commonly used techniques to handle imbalanced data is re-sampling of data as it is independent of the classifier being used. In this paper, we develop a hybrid re-sampling approach named SOS-BUS by combining well known oversampling technique SMOTE with our novel under-sampling technique. Our methodology aims to focus on the necessary data of majority class and avoid their removal in order to overcome the limitation of random under-sampling. Experimental results show that the proposed approach outperforms the other reference techniques in terms of Area under ROC Curve (AUC).