A hybrid neural network classifier combining ordered fuzzy ARTMAP and the dynamic decay adjustment algorithm

This paper presents a novel conflict-resolving neural network classifier that combines the ordering algorithm, fuzzy ARTMAP (FAM), and the dynamic decay adjustment (DDA) algorithm, into a unified framework. The hybrid classifier, known as Ordered FAMDDA, applies the DDA algorithm to overcome the limitations of FAM and ordered FAM in achieving a good generalization/performance. Prior to network learning, the ordering algorithm is first used to identify a fixed order of training patterns. The main aim is to reduce and/or avoid the formation of overlapping prototypes of different classes in FAM during learning. However, the effectiveness of the ordering algorithm in resolving overlapping prototypes of different classes is compromised when dealing with complex datasets. Ordered FAMDDA not only is able to determine a fixed order of training patterns for yielding good generalization, but also is able to reduce/resolve overlapping regions of different classes in the feature space for minimizing misclassification during the network learning phase. To illustrate the effectiveness of Ordered FAMDDA, a total of ten benchmark datasets are experimented. The results are analyzed and compared with those from FAM and Ordered FAM. The outcomes demonstrate that Ordered FAMDDA, in general, outperforms FAM and Ordered FAM in tackling pattern classification problems.     

The MSFAM: a modified fuzzy ARTMAP system

A fuzzy ARTMAP system is a system for incremental supervised learning of recognition categories and multi-dimensional maps in response to an arbitrary sequence of analog or binary input vectors. Fuzzy ARTMAP systems have been benchmarked against a variety of machine learning, neural networks, and genetic algorithms with considerable success. Owing to many appealing properties, fuzzy ARTMAP systems provide a natural basis for many researchers. Many different approaches have been proposed to modify fuzzy ARTMAP systems. In this paper, we propose a new approach to modifying a fuzzy ARTMAP system. We refer to the new system as the modified and simplified fuzzy ARTMAP (MSFAM) system. The aims of MSFAM systems are not only to reduce the architectural redundancy of the fuzzy ARTMAP system, but also to make extracted rules more comprehensible and concise. Four data sets were used for demonstrating the performance of the proposed MSFAM system.     

Fuzzy semi-parametric partially linear model with fuzzy inputs and fuzzy outputs

A large number of accounting studies have focused on parametric or non-parametric forms of fuzzy regression relationships between dependent and independent variables. Notably, semi-parametric partially linear model as a powerful tool to incorporate statistical parametric and non-parametric regression analyses has gained attentions in many real-life applications recently. However, fuzzy data find application in many real studies. This study is an investigation of semi-parametric partially linear model for such cases to improve the conventional fuzzy linear regression models with fuzzy inputs, fuzzy outputs, fuzzy smooth function and non-fuzzy coefficients. For this purpose, a hybrid procedure is suggested based on curve fitting methods and least absolutes deviations to estimate the fuzzy smooth function and fuzzy coefficients. The proposed method is also examined to be compared with a common fuzzy linear regression model via a simulation data set and some real fuzzy data sets. It is shown that the proposed fuzzy regression model performs more convenient and efficient results in regard to six goodness-of-fit criteria which concludes that the proposed model could be a rational substituted model of some common fuzzy regression models in many practical studies of fuzzy regression model in expert and intelligent systems.     

Fuzzy model tuning based on a training set with fuzzy model output values

The paper considers the training of a fuzzy model with the help of a training set with fuzzy model output values. Two ways are proposed for constructing fuzzy rule-based multifactor models that produce fuzzy numbers at their outputs. The problem of tuning such fuzzy models on the basis of a fuzzy training set is formulated, methods of its solution are considered, and relevant examples are presented. Computational experiments showed that training based on fuzzy data improves the modeling accuracy for both crisp and fuzzy test sets. __________ Translated from Kibernetika i Sistemnyi Analiz, No. 3, pp. 26–32, May–June 2007.     

GOFAM: a hybrid neural network classifier combining fuzzy ARTMAP and genetic algorithm

Fuzzy ARTMAP (FAM), which is a supervised model from the adaptive resonance theory (ART) neural network family, is one of the conspicuous neural network classifier. The generalization/performance of FAM is affected by two important factors which are network parameters and presentation order of training data. In this paper we introduce a genetic algorithm to find a better presentation order of training data for FAM. The proposed method which is the combination of genetic algorithm with Fuzzy ARTMAP is called Genetic Ordered Fuzzy ARTMAP (GOFAM). To illustrate the effectiveness of GOFAM, several standard datasets from UCI repository of machine learning databases are experimented. The results are analyzed and compared with those from FAM and Ordered FAM which is used to determine a fixed order of training pattern presentation to FAM. Experimental results demonstrate the performance of GOFAM is much better than performance of Fuzzy ARTMAP and Ordered Fuzzy ARTMAP. In term of network size, GOFAM performs significantly better than FAM and Ordered FAM.     

A dynamic fuzzy model for a drum-boiler-turbine system

A nonlinear dynamic fuzzy model for natural circulation drum-boiler-turbine is presented. The model is derived from Åström-Bell nonlinear dynamic system and describes the complicated dynamics of the physical plant. It is shown that the dynamic fuzzy model gives in some appropriate sense accurate global nonlinear prediction and at the same time that its local models are close approximations to the local linearizations of the nonlinear dynamic system. This closeness is illustrated by simulation in various conditions.     

The facility layout problem approached using a fuzzy model and a genetic search

The problem of facility layout design is discussed, taking into account the uncertainty of production scenarios and the finite production capacity of the departments. The uncertain production demand is modelled by a fuzzy number, and constrained arithmetic operators are used in order to calculate the fuzzy material handling costs. By using a ranking criterion, the layout that represents the minimum fuzzy cost is selected. A flexible bay structure is adopted as a physical model of the system while an effective genetic algorithm is implemented to search for a near optimal solution in a fuzzy contest. Constraints on the aspect ratio of the departments are taken into account using a penalty function introduced into the fitness function of the genetic algorithm. The efficiency of the genetic algorithm proposed is tested in a deterministic context and the possibility of applying the fuzzy approach to a medium-large layout problem is explored.This revised version was published in June 2005 with corrected page numbers.     

An improved fuzzy neural network based on T–S model

An improved fuzzy neural network based on Takagi–Sugeno (T–S) model is proposed in this paper. According to characteristics of samples spatial distribution the number of linguistic values of every input and the means and deviations of corresponding membership functions are determined. So the reasonable fuzzy space partition is got. Further a subtractive clustering algorithm is used to derive cluster centers from samples. With the parameters of linguistic values the cluster centers are fuzzified to get a more concise rule set with importance for every rule. Thus redundant rules in the fuzzy space are deleted. Then antecedent parts of all rules determine how a fuzzification layer and an inference layer connect. Next, weights of the defuzzification layer are initialized by a least square algorithm. After the network is built, a hybrid method combining a gradient descent algorithm and a least square algorithm is applied to tune the parameters in it. Simultaneous, an adaptive learning rate which is identified from input-state stability theory is adopted to insure stability of the network. The improved T–S fuzzy neural network (ITSFNN) has a compact structure, high training speed, good simulation precision, and generalization ability. To evaluate the performance of the ITSFNN, we experiment with two nonlinear examples. A comparative analysis reveals the proposed T–S fuzzy neural network exhibits a higher accuracy and better generalization ability than ordinary T–S fuzzy neural network. Finally, it is applied to predict markup percent of the construction bidding system and has a better prediction capability in comparison to some previous models.     

An improved dynamic modeling of a multibody system with spherical joints

A dynamic modeling of multibody systems having spherical joints is reported in this work. In general, three intersecting orthogonal revolute joints are substituted for a spherical joint with vanishing lengths of intermediate links between the revolute joints. This procedure increases sizes of associated matrices in the equations of motion, thus increasing computational burden of an algorithm used for dynamic simulation and control. In the proposed methodology, Euler parameters, which are typically used for representation of a rigid-body orientation in three-dimensional Cartesian space, are employed to represent the orientation of a spherical joint that connects a link to its previous one providing three-degree-of-freedom motion capability. For the dynamic modeling, the concept of the Decoupled Natural Orthogonal Complement (DeNOC) matrices is utilized. It is shown in this work that the representation of spherical joints motion using Euler parameters avoids the unnecessary introduction of the intermediate links, thereby no increase in the sizes of the associated matrices with the dynamic equations of motion. To confirm the efficiency of the proposed representation, it is illustrated with the dynamic modeling of a spatial four-bar Revolute-Spherical–Spherical-Revolute (RSSR) mechanism, where the CPU time of the dynamic modeling based on proposed methodology is compared with that based on the revolute joints substitution. Finally, it is explained how a complex suspension and steering linkage can be modeled using the proposed concept of Euler parameters to represent a spherical joint.     

Application of a modified fuzzy ARTMAP with feature-weight learning for the fault diagnosis of bearing

Considering different importance of the feature parameters to the fault conditions of bearing, a modified fuzzy ARTMAP (FAM) network model based on the feature-weight learning is presented in this paper. The features in time-domain, frequency-domain and wavelet-domain are extracted from the vibration signals to characterize the information relevant to the fault conditions of bearing. By the improved distance evaluation technique the optimal features are selected and the corresponding feature-weights which are assigned to the features to indicate their different importance to the fault conditions of bearing are obtained. Then they are combined with the modified FAM which is described by the weighted Manhattan distance and applied to the seven-class fault diagnosis of bearing. To assess the effectiveness and stability of the modified FAM network, bootstrapping method is employed to quantify the stability of the network performance statistically. Diagnosis results show that the modified FAM can more reliably and accurately recognize different fault classes.     

An improved grey dynamic GM(2, 1) model

In recent years, the grey dynamic model GM(2, 1), which is based on the grey system theory has been successfully applied in many prediction fields to solve efficiently the predicted problems of uncertainty systems. However, this model has irrational problem concerning the calculation of initial value and derivative. Therefore, the predicted accuracy of GM(2, 1) is unsatisfying when original data show great randomness. In this paper, the new calculation of initial value and derivative are first proposed to enhance the predicted power according to cubic spline function. The newly generated model is defined as 3spGM(2, 1). To further improve predicted accuracy, the Taylor approximation method is then applied to 3spGM(2, 1). We call the improved version as T-3spGM(2, 1). Finally, the effectiveness of proposed model is validated by using fault data sets of electric product.     

TPF: a dynamic system thrashing protection facility

Operating system designers attempt to keep high CPU utilization by maintaining an optimal multiprogramming level (MPL). Although running more processes makes it less likely to leave the CPU idle, too many processes adversely incur serious memory competition, and even introduce thrashing, which eventually lowers CPU utilization. A common practice to address the problem is to lower the MPL with the aid of process swapping out/in operations. This approach is expensive and is only used when the system begins serious thrashing. The objective of our study is to provide highly responsive and cost‐effective thrashing protection by adaptively conducting priority page replacement in a timely manner. We have designed a dynamic system Thrashing Protection Facility (TPF) in the system kernel. Once TPF detects system thrashing, one of the active processes will be identified for protection. The identified process will have a short period of privilege in which it does not contribute its least recently used (LRU) pages for removal so that the process can quickly establish its working set, improving the CPU utilization. With the support of TPF, thrashing can be eliminated in its early stage by adaptive page replacement, so that process swapping will be avoided or delayed until it is truly necessary. We have implemented TPF in a current and representative Linux kernel running on an Intel Pentium machine. Compared with the original Linux page replacement, we showthat TPF consistently and significantly reduces page faults and the execution time of each individual job in several groups of interacting SPEC CPU2000 programs. We also show that TPF introduces little additional overhead to program executions, and its implementation in Linux (or Unix) systems is straightforward. Copyright © 2002 John Wiley & Sons, Ltd.     

An improved robust fuzzy-PID controller with optimal fuzzy reasoning.

Many fuzzy control schemes used in industrial practice today are based on some simplified fuzzy reasoning methods, which are simple but at the expense of losing robustness, missing fuzzy characteristics, and having inconsistent inference. The concept of optimal fuzzy reasoning is introduced in this paper to overcome these shortcomings. The main advantage is that an integration of the optimal fuzzy reasoning with a PID control structure will generate a new type of fuzzy-PID control schemes with inherent optimal-tuning features for both local optimal performance and global tracking robustness. This new fuzzy-PID controller is then analyzed quantitatively and compared with other existing fuzzy-PID control methods. Both analytical and numerical studies clearly show the improved robustness of the new fuzzy-PID controller.     

An improved ranking method for fuzzy numbers with integral values

Ranking fuzzy numbers is a very important decision-making procedure in decision analysis and applications. The last few decades have seen a large number of approaches investigated for ranking fuzzy numbers, yet some of these approaches are non-intuitive and inconsistent. In 1992, Liou and Wang proposed an approach to rank fuzzy number based a convex combination of the right and the left integral values through an index of optimism. Despite its merits, some shortcomings associated with Liou and Wang's approach include: (i) it cannot differentiate normal and non-normal fuzzy numbers, (ii) it cannot rank effectively the fuzzy numbers that have a compensation of areas, (iii) when the left or right integral values of the fuzzy numbers are zero, the index of optimism has no effect in either the left integral value or the right integral value of the fuzzy number, and (iv) it cannot rank consistently the fuzzy numbers and their images.This paper proposes a revised ranking approach to overcome the shortcomings of Liou and Wang's ranking approach. The proposed ranking approach presents the novel left, right, and total integral values of the fuzzy numbers. The median value ranking approach is further applied to differentiate fuzzy numbers that have the compensation of areas. Finally, several comparative examples and an application for market segment evaluation are given herein to demonstrate the usages and advantages of the proposed ranking method for fuzzy numbers.     

Approximate dynamic programming with a fuzzy parameterization

Dynamic programming (DP) is a powerful paradigm for general, nonlinear optimal control. Computing exact DP solutions is in general only possible when the process states and the control actions take values in a small discrete set. In practice, it is necessary to approximate the solutions. Therefore, we propose an algorithm for approximate DP that relies on a fuzzy partition of the state space, and on a discretization of the action space. This fuzzy Q-iteration algorithm works for deterministic processes, under the discounted return criterion. We prove that fuzzy Q-iteration asymptotically converges to a solution that lies within a bound of the optimal solution. A bound on the suboptimality of the solution obtained in a finite number of iterations is also derived. Under continuity assumptions on the dynamics and on the reward function, we show that fuzzy Q-iteration is consistent, i.e., that it asymptotically obtains the optimal solution as the approximation accuracy increases. These properties hold both when the parameters of the approximator are updated in a synchronous fashion, and when they are updated asynchronously. The asynchronous algorithm is proven to converge at least as fast as the synchronous one. The performance of fuzzy Q-iteration is illustrated in a two-link manipulator control problem.     

EFM: evolutionary fuzzy model for dynamic activities recognition using a smartphone accelerometer

Activity recognition is an emerging field of research that enables a large number of human-centric applications in the u-healthcare domain. Currently, there are major challenges facing this field, including creating devices that are unobtrusive and handling uncertainties associated with dynamic activities. In this paper, we propose a novel Evolutionary Fuzzy Model (EFM) to measure the uncertainties associated with dynamic activities and relax the domain knowledge constraints which are imposed by domain experts during the development of fuzzy systems. Based on the time and frequency domain features, we define the fuzzy sets and estimate the natural grouping of data through expectation maximization of the likelihoods. A Genetic Algorithm (GA) is investigated and designed to determine the optimal fuzzy rules. To evaluate the EFM, we performed experiments on seven daily life activities of ten human subjects. Our experiments show significant improvement of 9 % in class-accuracy and 11 % in the F-measures of recognized activities compared to existing counterparts. The practical solution to dynamic activity recognition problems is expected to be an EFM, due to EFM’s utilization of smartphones and natural way of handling uncertainties.     

Fuzzy SVM with a new fuzzy membership function

It is known that with a proper fuzzy membership function, a fuzzy support vector machine can effectively reduce the effects of outliers when solving the classification problem. In this paper, a new fuzzy membership function is proposed to the nonlinear fuzzy support vector machine. The fuzzy membership is calculated in the feature space and is represented by kernels. This method gives good performance on reducing the effects of outliers and significantly improves the classification accuracy and generalization.