Dynamic and Static Weighting in Classifier Fusion

Valdovinos, R. M.; Sánchez, J. Salvador; Barandela, Ricardo

doi:10.1007/11492542_8

Cited by 19 publications

(11 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If the classifier outputs are interpreted as fuzzy membership values, fuzzy approaches could be used. It is also possible to train the combiner separately, using outputs of the base classifiers as new features [6]. One of the earliest works on MCS dates back to Dasarathy and Sheela's 1979 work, which discussed the idea of partitioning the feature space using two or more classifiers [7].…”

Section: Fig 1 Hierarchy Of Mcs Fusion Methodsmentioning

confidence: 99%

Making Diversity Enhancement Based on Multiple Classifier System by Weight Tuning

Haghighi

Vahedian

Yazdi

2011

Neural Process Lett

View full text Add to dashboard Cite

This article presents a new method to construct multiple classifier system by making diverse base classifiers using weight tuning. In the method presented, base classifiers are multilayer perceptions which creates diverse base classifiers using a three-step procedure. In the first step, base classifiers are trained for acceptable accuracy. In the second step, a weight tuning process tunes their weights such that each one can distinguish one class of input data from the others with highest possible accuracy. An evolutionary method is used to optimize efficiency of each base classifier to distinguish one class of input data in this step. In the third step, a new method combines the results of the base classifiers. As diversity is measured and monitored throughout the entire procedure, it is measured using a confusion matrix. Superiority of the proposed method is discussed using several known classifier fusion methods and known benchmark datasets.

show abstract

Section: Fig 1 Hierarchy Of Mcs Fusion Methodsmentioning

confidence: 99%

Making Diversity Enhancement Based on Multiple Classifier System by Weight Tuning

Haghighi

Vahedian

Yazdi

2011

Neural Process Lett

View full text Add to dashboard Cite

show abstract

“…A weighted k-NN rule for classifying new patterns was first proposed by Dudani [16]. The votes of the k nearest neighbors are weighted by a function of their distance to the test pattern.…”

Section: Fuzzy Match Score Of Semantic Service Matchmentioning

confidence: 99%

Fuzzy match score of semantic service match

Ludwig

2008

2008 IEEE International Conference on Fuzzy Systems (IEEE World Congress on Computational Intelligence)

View full text Add to dashboard Cite

Automatic discovery of services is a crucial task for the e-Science and e-Business communities. Finding a suitable way to address this issue has become one of the key points to convert the Web into a distributed source of computation, as it enables the location of distributed services to perform a required functionality. To provide such an automatic location, the discovery process should be based on the semantic match between a declarative description of the service being sought and a description being offered. This problem requires not only an algorithm to match these descriptions, but also a language to declaratively express the capabilities of services. The proposed matchmaking approach is based on semantic descriptions for service attributes, descriptions and metadata. For the ranking of service matches a match score is calculated whereby the weight values are either given by the user or estimated using a fuzzy approach. An evaluation of both weight assignment approaches is conducted identifying in which scenario one works better than the other.

show abstract

“…The input patterns are partitioned, and the best classifier is nominated for each partition. While a static fusion method employs constant weights for each classifier based on training, a dynamic fusion method changes the weights of each classifier based on the observed test pattern [12]. For example, the distance of a test pattern to its nearest neighbor for each individual classifier may be used to compute the dynamic weights.…”

Section: Previous Researchmentioning

confidence: 99%

“…The process involves selecting the minimum value in each column of the DP matrix, and then declaring a classifier with the maximum value of these minimum values in each column as the best classifier. This method is used in [12] to classify meeting events.…”

Section: Previous Researchmentioning

confidence: 99%

Dynamic fusion of classifiers for fault diagnosis

Singh

Choi

Kodali

et al. 2007

2007 IEEE International Conference on Systems, Man and Cybernetics

View full text Add to dashboard Cite

Abstract-This paper considers the problem of temporally fusing classifier outputs to improve the overall diagnostic classification accuracy in safety-critical systems. Here, we discuss dynamic fusion of classifiers which is a special case of the dynamic multiple fault diagnosis (DMFD) problem [1]-[3]. The DMFD problem is formulated as a maximum a posteriori (MAP) configuration problem in tri-partite graphical models, which is NP-hard. A primal-dual optimization framework is applied to solve the MAP problem. Our process for dynamic fusion consists of four key steps: (1) data preprocessing such as noise suppression, data reduction and feature selection using data-driven techniques, (2) error correcting codes to transform the multiclass data into binary classification, (3) fault detection using pattern recognition techniques (support vector machines in this paper), and (4) dynamic fusion of classifiers output labels over time using the DMFD algorithm. An automobile engine data set, simulated under various fault conditions [4], was used to illustrate the fusion process. The results demonstrate that an ensemble of classifiers, when fused over time, reduces the diagnostic error as compared to a single classifier and static fusion of classifiers trained over the entire batch of data. The results for sliding window dynamic fusion are also provided.

show abstract

Dynamic and Static Weighting in Classifier Fusion

Cited by 19 publications

References 11 publications

Making Diversity Enhancement Based on Multiple Classifier System by Weight Tuning

Making Diversity Enhancement Based on Multiple Classifier System by Weight Tuning

Fuzzy match score of semantic service match

Dynamic fusion of classifiers for fault diagnosis

Contact Info

Product

Resources

About