A family of measures for best top- n class-selective decision rules

When classes strongly overlap in the feature space, or when some classes are not known in advance, the performance of a classifier heavily decreases. To overcome this problem, the reject option has been introduced. It simply consists in withdrawing the decision, and let another classifier, or an exp...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Pattern recognition Ročník 45; číslo 1; s. 552 - 562
Hlavní autoři: Le Capitaine, Hoel, Frélicot, Carl
Médium: Journal Article
Jazyk:angličtina
Vydáno: Kidlington Elsevier Ltd 2012
Elsevier
Témata:
Applied sciences
Class-selective decision rules
Classification
Classifiers
Computer Science
Computer Vision and Pattern Recognition
Exact sciences and technology
Fuzzy aggregation operators
Information, signal and communications theory
Intervals
Labels
Operators
Pattern recognition
Public domain
Reject options
Signal and communications theory
Signal representation. Spectral analysis
Signal, noise
Telecommunications and information theory
Thresholds
Class-selective decision rules
Fuzzy aggregation operators
Reject options
Performance evaluation
Automatic classification
Signal classification
Decision rule
ISSN:0031-3203, 1873-5142
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:When classes strongly overlap in the feature space, or when some classes are not known in advance, the performance of a classifier heavily decreases. To overcome this problem, the reject option has been introduced. It simply consists in withdrawing the decision, and let another classifier, or an expert, take the decision whenever exclusively classifying is not reliable enough. The classification problem is then a matter of class-selection, from none to all classes. In this paper, we propose a family of measures suitable to define such decision rules. It is based on a new family of operators that are able to detect blocks of similar values within a set of numbers in the unit interval, the soft labels of an incoming pattern to be classified, using a single threshold. Experiments on synthetic and real datasets available in the public domain show the efficiency of our approach. ► We propose a family of measures suitable to define class-selective decision rules. ► The measures are based on a block-similarity detection of ordered membership degrees. ► Degrees are combined thanks to fuzzy integrals, where weights are defined by kernels. ► Compared to usual rules, the approach is shown to be efficient on benchmark data.
Bibliografie:ObjectType-Article-2
SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 23
ISSN:0031-3203
1873-5142
DOI:10.1016/j.patcog.2011.04.030