Many functionals have been proposed for validation of partitions object data produced by the fuzzy c-means (FCM) clustering algorithm. We examine role a subtle but important parameter-the weighting exponent m FCM model-plays in determining validity partitions. The considered are partition coefficient and entropy indexes Bezdek, Xie-Beni (1991), extended indexes, Fukuyama-Sugeno index (1989). Limit analysis indicates, numerical experiments confirm, that is sensitive to both high low values...

In 1997, we proposed the fuzzy-possibilistic c-means (FPCM) model and algorithm that generated both membership typicality values when clustering unlabeled data. FPCM constrains so sum over all data points of typicalities to a cluster is one. The row constraint produces unrealistic for large sets. this paper, propose new called possibilistic-fuzzy (PFCM) model. PFCM memberships possibilities simultaneously, along with usual point prototypes or centers each cluster. hybridization possibilistic...

We review two clustering algorithms (hard c-means and single linkage) three indexes of crisp cluster validity (Hubert's statistics, the Davies-Bouldin index, Dunn's index). illustrate deficiencies index which make it overly sensitive to noisy clusters propose several generalizations that are not as brittle outliers in clusters. Our numerical examples show standard measure interset distance (the minimum between points a pair sets) is worst (least reliable) upon base validation when expected...

Proposes a simple but robust model independent self-tuning scheme for fuzzy logic controllers (FLCs). Here, the output scaling factor (SF) is adjusted online by rules according to current trend of controlled process. The rule-base tuning SF defined on error (e) and change (/spl Delta/e) variable using most natural unbiased membership functions (MFs). proposed technique applied both PI- PD-type FLCs conduct simulation analysis wide range different linear nonlinear second-order processes...

The relationship between the sequential hard c-means (SHCM) and learning vector quantization (LVQ) clustering algorithms is discussed. impact interaction of these two families methods with Kohonen's self-organizing feature mapping (SOFM), which not a method but often lends ideas to algorithms, are considered. A generalization LVQ that updates all nodes for given input proposed. network attempts find minimum well-defined objective function. rules depend on degree distance match winner node;...

This paper presents an online feature selection algorithm using genetic programming (GP). The proposed GP methodology simultaneously selects a good subset of features and constructs classifier the selected features. For c-class problem, it provides having c trees. In this context, we introduce two new crossover operations to suit process. As byproduct, our produces ranking scheme. We tested method on several data sets dimensions varying from 4 7129. compared performance with results...

We present an integrated algorithm for simultaneous feature selection (FS) and designing of diverse classifiers using a steady state multiobjective genetic programming (GP), which minimizes three objectives: 1) false positives (FPs); 2) negatives (FNs); 3) the number leaf nodes in tree. Our method divides c -class problem into binary classification problems. It evolves sets programs to create ensembles. During mutation operation, our exploits fitness as well unfitness features, dynamically...

We propose an embedded/integrated feature selection method based on neural networks with Group Lasso penalty. regularization is considered to produce sparsity the inputs network, i.e., for of useful features. using a multi-layer perceptron usually requires additional set weights, while our formulation does not require that. However, penalty non-differentiable at origin. This may lead oscillations in numerical simulations and make it difficult analyze theoretically. To address this issue,...

A major limitation of fuzzy or neuro-fuzzy systems is their failure to deal with high-dimensional datasets. This happens primarily due the use T-norm, particularly, product minimum (or a softer version it). Thus, there are hardly any work dealing datasets having features more than hundred so. Here, we propose framework that can handle even 7000 features! In this context, an adaptive softmin (Ada-softmin) which effectively overcomes drawbacks "numeric underflow" and "fake minimum" arise for...

We justify the need for computing both membership and typicality values when clustering unlabeled data. Then we propose a new model called fuzzy-possibilistic c-means (FPCM). Unlike fuzzy possibilistic (FCM/PCM) models, FPCM simultaneously produces memberships possibilities, along with usual point prototypes or cluster centers each show that solves noise sensitivity defect of FCM, also overcomes coincident clusters problem PCM. derive first order necessary conditions extrema PFCM objective...

Shannon's definition of entropy is critically examined and a new classical based on the exponential behavior information gain proposed along with its justification. The concept extended to defining global, local, conditional gray-level image. Based these definitions four algorithms for object extraction are developed. One uses Poisson distribution-based model an ideal A positional giving any regarding location in scene introduced.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML"...

First, this paper reviews several well known measures of fuzziness for discrete fuzzy sets. Then new multiplicative and additive classes are defined. We show that each class satisfies five well-known axioms measures, demonstrate existing relatives these classes. The is based on nonnegative, monotone increasing concave functions. requires only nonnegative Some relationships between the established, some properties derived. relative merits drawbacks different applications discussed. A weighted...

The definition of Shannon&apos;s entropy in the context information theory is critically examined and some its applications to image processing problems are reviewed. A new classical based on exponential behaviour information-gain proposed along with justification. Its properties also include those entropy. concept then extended fuzzy sets for defining a non-probabilistic grey tone global, local conditional Based definitions, three algorithms developed segmentation. superiority these...

The authors propose a fuzzy Kohonen clustering network which integrates the c-means (FCM) model into learning rate and updating strategies of network. This yields an optimization problem related to FCM, numerical results show improved convergence as well reduced labeling errors. It is proved that proposed scheme equivalent algorithms. new method can be viewed type but it self-organizing, since size update neighborhood in competitive layer are automatically adjusted during learning....

Most methods of classification either ignore feature analysis or do it in a separate phase, offline prior to the main task. This paper proposes neuro-fuzzy scheme for designing classifier along with selection. It is four-layered feed-forward network realizing fuzzy rule-based classifier. The trained by error backpropagation three phases. In first learns important features and rules. subsequent phases, pruned an "optimal" architecture that represents set Pruning found drastically reduce size...

We propose a new approach for designing classifiers c-class (c/spl ges/2) problem using genetic programming (GP). The proposed takes an integrated view of all classes when the GP evolves. A multitree representation chromosomes is used. In this context, we modified crossover operation and mutation that reduces destructive nature conventional operations. use concept unfitness tree to select trees This gives more opportunity unfit become fit. OR-ing in terminal population introduced, which...

Facial expressions of a person representing similar emotion are not always unique. Naturally, the facial features subject taken from different instances same have wide variations. In presence two or more features, variation attributes together makes recognition problem complicated. This is main source uncertainty in problem, which has been addressed here steps using type-2 fuzzy sets. First face space constructed with background knowledge subjects for emotions. Second, an unknown expression...

Characterization of dissimilarity/divergence between intuitionistic fuzzy sets (IFSs) is important as it has applications in different areas including image segmentation and decision making. This study deals with the problem comparison sets. An axiomatic definition divergence measures for IFSs presented, which are particular cases dissimilarities IFSs. The relationships among IF-divergences, IF-dissimilarities, IF-distances studied. Finally, we propose a very general framework IFSs, where...