Results 1  10
of
13
Robust image segmentation using FCM with spatial constraints based on new kernelinduced distance measure
 IEEE Transactions on Systems, Man and Cybernetics, Part B
, 2004
"... Abstract Fuzzy cmeans clustering (FCM) with spatial constraints (FCM_S) is an effective algorithm suitable for image segmentation. Its effectiveness contributes not only to introduction of fuzziness for belongingness of each pixel but also to exploitation of spatial contextual information. Altho ..."
Abstract

Cited by 68 (5 self)
 Add to MetaCart
(Show Context)
Abstract Fuzzy cmeans clustering (FCM) with spatial constraints (FCM_S) is an effective algorithm suitable for image segmentation. Its effectiveness contributes not only to introduction of fuzziness for belongingness of each pixel but also to exploitation of spatial contextual information. Although the contextual information can raise its insensitivity to noise to some extent, FCM_S (1) still lacks enough robustness to noise and outliers and (2) is not suitable for revealing nonEuclidean structure of the input data due to the use of Euclidean distance (L2 norm). In this paper, to overcome the above problems, we first propose two variants, FCM_S1 and FCM_S2, of FCM_S to aim at simplifying its computation and then extend them, including
A novel kernelized fuzzy cmeans algorithm with application in medical image segmentation
 Artificial Intelligence in Medicine
, 2004
"... image segmentation ..."
(Show Context)
Graph nodes clustering with the sigmoid commutetime kernel: A . . .
 DATA & KNOWLEDGE ENGINEERING
, 2009
"... ..."
Automated FuzzyClustering for DoctuS Expert System
"... 'Doctus ' 1 is capable of deduction also called rulebased reasoning and of induction, which is the symbolic version of reasoning by cases 2. If connected to databases or data warehouses the inductive reasoning of Doctus is also used for data mining. To handle numerical domains Doctus uses ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
(Show Context)
'Doctus ' 1 is capable of deduction also called rulebased reasoning and of induction, which is the symbolic version of reasoning by cases 2. If connected to databases or data warehouses the inductive reasoning of Doctus is also used for data mining. To handle numerical domains Doctus uses statistical clustering algorithm. We define the problem in three steps: how to perform a clustering, which is neither rigid nor sensitive to noise, benefiting from the properties of the application domain, reducing the complexity as much as possible, and supplying the decision maker with useful information enabling the possibility of interaction? In this paper we present the conception of Automated FuzzyClustering using triangular and trapezoidal Fuzzysets, which provides overlapping Fuzzyset covering of the domain. I. FUZZY CLUSTERING FOR SYMBOLIC ES – WHY? We investigate the expert systems in supporting the business decision making process. Let’s first examine the domain of the application, to map characteristics that are important to choose the appropriate tool for support. We are dealing with decision making of a leader and of a manager on the expert level of knowledge and higher, who are to considering much of soft information and hard data, and use heuristic processes to take the decisions. First there is a need to discover the properties of the heuristic processes in comparison to other processes: 1. At deterministic processes there is an expected value only with no dispersion. It is determined what output follows a particular input, it will happen in 100 % of repetitions. Small changes on the input will result in small changes on the output, which can be calculated precisely. Deterministic processes can be met e.g. in classical physics (mechanics of not microscopic, but also not astronomy sized bodies). 2. Output of a stochastic process can be described with its expected value and its dispersion, which is smaller at least one order of magnitude. Small changes on the input will result in small changes on the output, which can be 1 www.doctus.info 2 Originally
$\varepsilon$Insensitive Learning Techniques for Approximate Reasoning Systems
 JOUR. OF COMPUTATIONAL COGNITION
, 2003
"... Initially, an axiomatic approach to the definitions of fuzzy connectives are recalled. Based on these definitions several important fuzzy connectives and their properties are described. Then, the idea of approximate reasoning using generalized modus ponens and fuzzy implication are considered. After ..."
Abstract
 Add to MetaCart
Initially, an axiomatic approach to the definitions of fuzzy connectives are recalled. Based on these definitions several important fuzzy connectives and their properties are described. Then, the idea of approximate reasoning using generalized modus ponens and fuzzy implication are considered. After reviewing the wellknown fuzzy systems, the artificial neural network based on logical interpretation of ifthen rules is presented. The elimination of noninformative part of final fuzzy set before defuzzification plays the key role in this system. Then, new learning methods tolerant to imprecision are introduced and used to learning this system. The proposed learning methods make it possible to dispose an intrinsic inconsistency of neurofuzzy modeling, where zerotolerance learning is used to obtain fuzzy model tolerant to imprecision. These new methods may be called #insensitive learning, where, in order to fit the fuzzy model to real data, the weighted #insensitive loss function is used. The #insensitive learning leads to a fuzzy model with minimal VapnikChervonenkis dimension, which results in an improved generalization ability of this system. Another advantage of the proposed learning methods is their outliers robustness. In this paper, two approaches to solving the #insensitive learning problem are presented. The first approach leads to a quadratic programming problem with bound constraints and one linear equality constraint. The second approach leads to a problem of solving a system of linear inequalities. Three computationally efficient numerical methods for the #insensitive learning are proposed. Finally, an example is given to demonstrate the validity of the introduced methods.
SOCA DOI 10.1007/s117610110090y SPECIAL ISSUE PAPER
"... Weighted fuzzy clustering for capabilitydriven service aggregation ..."
(Show Context)
AN EFFICIENT ALGORITHM FOR CLUSTERING OF IMAGES USING FUZZY LOCAL INFORMATION C
"... Fuzzy CMeans (FCM) is hardclustering algorithms. It is effective for spherical clusters, it does not perform well for more general clusters and FCM algorithm has also been extended to the kernel FCM algorithm, which yields better performance. Multiple Kernel Fuzzy CMeans (MKFC) algorithm simultan ..."
Abstract
 Add to MetaCart
Fuzzy CMeans (FCM) is hardclustering algorithms. It is effective for spherical clusters, it does not perform well for more general clusters and FCM algorithm has also been extended to the kernel FCM algorithm, which yields better performance. Multiple Kernel Fuzzy CMeans (MKFC) algorithm simultaneously find the best degrees of membership and the optimal kernel weights for a nonnegative combination of a set of kernels. But it takes more time for segmentation and it does not considered the noise pixel. To overcome this problem, Fuzzy Local Information CMeans (FLICM) algorithm is used. The major characteristic of FLICM is the use of a fuzzy local gray level similarity measure, aiming to guarantee noise insensitiveness and image detail preservation. The algorithm is fully free of the empirically adjusted parameters incorporated into all other fuzzy cmeans algorithms. FLICM technique reduces the time complexity at the same time managing the noise pixels in the image.
unknown title
"... dInst Acrosssubject variability ow n be l va e s irty eresu to t ustr tional networks observed using cognitive subtractions (e.g. those associated with aging s and se etwork f an infi n of vi of “cognitive subtraction ” that compares neuronal activation during a andNetwork 3 involvesfinger press r ..."
Abstract
 Add to MetaCart
(Show Context)
dInst Acrosssubject variability ow n be l va e s irty eresu to t ustr tional networks observed using cognitive subtractions (e.g. those associated with aging s and se etwork f an infi n of vi of “cognitive subtraction ” that compares neuronal activation during a andNetwork 3 involvesfinger press responses.We further assume that
Full article Image segmentation techniques in medical sciences
"... Abstract: Classical and clustering techniques for image segmentation are important tools in medical sciences. Classical techniques include histogram, region growing, watershed, and contour. The more recent clustering techniques include standard fuzzy cmeans clustering, kernelized cmeans, spatial c ..."
Abstract
 Add to MetaCart
(Show Context)
Abstract: Classical and clustering techniques for image segmentation are important tools in medical sciences. Classical techniques include histogram, region growing, watershed, and contour. The more recent clustering techniques include standard fuzzy cmeans clustering, kernelized cmeans, spatial constrained fuzzy cmeans, and kmeans clustering. These methods are applied on different images, synthetic image, T1weighted MR phantom, and real MR slices, which the performance of them are compared. The comparison is based on estimating the segmentation accuracy and time for each method when applied on three test images: synthetic image, T1weighted MR phantom, and real MR slices.