Credal c-means clustering method based on belief functions

Liu, Zhun-ga; Pan, Quan; Dezert, Jean; Mercier, Grégoire

doi:10.1016/j.knosys.2014.11.013

Cited by 123 publications

(71 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The main difference with FCM is that prototypes are defined not only for clusters, but also for sets of clusters (or "meta-clusters"). In [23], a variant of the ECM algorithm (called CCM) was proposed, based on an alternative definition of the distance between a vector and the prototype of a meta-cluster. This modification sometimes produces more sensible results in situations where the prototype of a meta-cluster is close to that of singleton cluster.…”

Section: Discussionmentioning

confidence: 99%

Evidential clustering of large dissimilarity data

Denźux

Sriboonchitta

Kanjanatarakul

2016

Knowledge-Based Systems

View full text Add to dashboard Cite

In evidential clustering, the membership of objects to clusters is considered to be uncertain and is represented by Dempster-Shafer mass functions, forming a credal partition. The EVCLUS algorithm constructs a credal partition in such a way that larger dissimilarities between objects correspond to higher degrees of conflict between the associated mass functions. In this paper, we present several improvements to EVCLUS, making it applicable to very large dissimilarity data. First, the gradient-based optimization procedure in the original EVCLUS algorithm is replaced by a much faster iterative row-wise quadratic programming method. Secondly, we show that EVCLUS can be provided with only a random sample of the dissimilarities, reducing the time and space complexity from quadratic to roughly linear. Finally, we introduce a two-step approach to construct credal partitions assigning masses to selected pairs of clusters, making the algorithm outputs more informative than those of the original EVCLUS, while remaining manageable for large numbers of clusters.

show abstract

Section: Discussionmentioning

confidence: 99%

Evidential clustering of large dissimilarity data

Denźux

Sriboonchitta

Kanjanatarakul

2016

Knowledge-Based Systems

View full text Add to dashboard Cite

show abstract

“…In some applications, class information is uncertain. This is the case, in particular, when the class labels cannot be observed directly and have to be inferred by an unsupervised learning algorithm (see, e.g., [18,28,29]) or by any other indirect method. Sometimes, class labels are assessed subjectively by an expert, a group of experts (as in the two examples described in Section 5.2), or even by a large number of individuals through crowdsourcing (see, e.g.…”

Section: Soft Labelsmentioning

confidence: 99%

Parametric classification with soft labels using the evidential EM algorithm: linear discriminant analysis versus logistic regression

Quost

Denœux

2017

Adv Data Anal Classif

View full text Add to dashboard Cite

classification with soft labels using the Evidential EM algorithm : linear discriminant analysis versus logistic regression.Abstract Partially supervised learning extends both supervised and unsupervised learning, by considering situations in which only partial information about the response variable is available. In this paper, we consider partially supervised classification and we assume the learning instances to be labeled by Dempster-Shafer mass functions, called soft labels. Linear discriminant analysis and logistic regression are considered as special cases of generative and discriminative parametric models. We show that the evidential EM algorithm can be particularized to fit the parameters in each of these models. We describe experimental results with simulated data sets as well as with two real applications: K-complex detection in sleep EEGs signals and facial expression recognition. These results confirm the interest of using soft labels for classification as compared to potentially erroneous crisp labels, when the true class membership is partially unknown or ill-defined.

show abstract

“…It has been widely employed in different fields [30,31]. Compared to the D-S theory of evidence, the calculation process of the ER rule is linear, which reduces the computational complexity [32]. Moreover, the most important thing is that it can deal with the conflict of evidence attributes which cannot be solved by using D-S theory [33].…”

Section: Introductionmentioning

confidence: 99%

Reliability Assessment of Cloud Computing Platform Based on Semiquantitative Information and Evidential Reasoning

Hang

Pei-li

2016

Journal of Control Science and Engineering

View full text Add to dashboard Cite

A reliability assessment method based on evidential reasoning (ER) rule and semiquantitative information is proposed in this paper, where a new reliability assessment architecture including four aspects with both quantitative data and qualitative knowledge is established. The assessment architecture is more objective in describing complex dynamic cloud computing environment than that in traditional method. In addition, the ER rule which has good performance for multiple attribute decision making problem is employed to integrate different types of the attributes in assessment architecture, which can obtain more accurate assessment results. The assessment results of the case study in an actual cloud computing platform verify the effectiveness and the advantage of the proposed method.

show abstract

Credal c-means clustering method based on belief functions

Cited by 123 publications

References 29 publications

Evidential clustering of large dissimilarity data

Evidential clustering of large dissimilarity data

Parametric classification with soft labels using the evidential EM algorithm: linear discriminant analysis versus logistic regression

Reliability Assessment of Cloud Computing Platform Based on Semiquantitative Information and Evidential Reasoning

Contact Info

Product

Resources

About