Fuzzy Rough Decision Trees for Multi-label Classification

Wang, Xiaoxue; An, Shuang; Shi, Hong; Hu, Qinghua

doi:10.1007/978-3-319-25783-9_19

Cited by 4 publications

(4 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Narayanan et al (2015) used three fuzzy partitioning techniques, namely fuzzy C-means clustering, grid partitioning and subtractive clustering, to induce fuzzy decision trees. Wang et al (2015b) proposed a multi-label decision tree algorithm based on fuzzy rough sets, which it could tackle with symbolic, continuous and fuzzy data. To improve classification accuracy and achieve faster convergence, a novel strategy, an ‘improved neuro fuzzy decision tree with an adaptive learning rate and momentum factor’, was considered by Narayanan et al (2016).…”

Section: Introductionmentioning

confidence: 99%

“…Moreover, based on the Map-Reduce programming model for generating both binary and multi-way trees from large data sets, a distributed fuzzy decision tree learning scheme was proposed by Segatori et al (2017). From the works of Bujnowski et al (2015), Chandra and Varghese (2008), Liu et al (2013), Manwani and Sastry (2012), Narayanan et al (2015, 2016), Segatori et al (2017), Umano et al (1994) and Wang et al (2000, 2015b), we know that the fuzzy decision trees can handle uncertainty and are easily accountable, but they cannot deal with the classification problem if the classification boundary is not parallel to the axis.…”

Section: Introductionmentioning

confidence: 99%

“…From the discussions of fuzzy decision trees (Bujnowski et al, 2015; Chandra and Varghese, 2008; Liu et al, 2013; Manwani and Sastry, 2012; Narayanan et al, 2015, 2016; Segatori et al, 2017; Umano et al, 1994;Wang et al, 2000, 2015b), oblique decision trees (Barros et al, 2014; Breiman et al, 1984; Cantu-Paz and Kamath, 2003; Kretowski, 2004; Li et al, 2003; Manwani and Sastry, 2012; Rivera-Lopez and Canul-Reich, 2017; Shah and Sastry, 1999; Wickramarachchi et al, 2016) and fuzzy rules decision trees (FRDTs) (Wang et al, 2015c), a FODT is proposed in this paper; its growth structure is depicted in Figure 1.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

New classification technique: fuzzy oblique decision tree

Cai

Sun

2018

Transactions of the Institute of Measurement and Control

View full text Add to dashboard Cite

Based on axiomatic fuzzy set (AFS) theory and fuzzy information entropy, a novel fuzzy oblique decision tree (FODT) algorithm is proposed in this paper. Traditional axis-parallel decision trees only consider a single feature at each non-leaf node, while oblique decision trees partition the feature space with an oblique hyperplane. By contrast, the FODT takes dynamic mining fuzzy rules as a decision function. The main idea of the FODT is to use these fuzzy rules to construct leaf nodes for each class in each layer of the tree; the samples that cannot be covered by the fuzzy rules are then put into an additional node – the only non-leaf node in this layer. Construction of the FODT consists of four major steps: (a) generation of fuzzy membership functions automatically by AFS theory according to the raw data distribution; (b) extraction of dynamically fuzzy rules in each non-leaf node by the fuzzy rule extraction algorithm (FREA); (c) construction of the FODT by the fuzzy rules obtained from step (b); and (d) determination of the optimal threshold δ to generate a final tree. Compared with five traditional decision trees (C4.5, LADtree (LAD), Best-first tree (BFT), SimpleCart (SC) and NBTree (NBT)) and a recently obtained fuzzy rules decision tree (FRDT) on eight UCI machine learning data sets and one biomedical data set (ALLAML), the experimental results demonstrate that the proposed algorithm outperforms the other decision trees in both classification accuracy and tree size.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

New classification technique: fuzzy oblique decision tree

Cai

Sun

2018

Transactions of the Institute of Measurement and Control

View full text Add to dashboard Cite

show abstract

“…Many mono-label classifiers were adapted to handle multilabel data, such as the K-Nearest Neighbours algorithm [3] [4], the naive bayes algorithm [5], the Support Vector machines [6] [7], the decision trees [8] [9], and the neural networks [10] [11]. In this paper we are more interested in transformation methods [12] because they can be used with any mono-label classifier.…”

Section: Introductionmentioning

confidence: 99%

Graded multi-label classification: Compromise between handling label relations and limiting error propagation

Laghmari¹,

Marsala²,

Ramdani³

2016

2016 11th International Conference on Intelligent Systems: Theories and Applications (SITA)

View full text Add to dashboard Cite

In graded multi-label classification (GMLC), each data can be assigned to multiple labels according to a degree of membership on an ordinal scale, and with respect to label relations. For example, in a movie catalog web page, a five stars action movie should be at least a one star suspense movie. Ignoring those relations can lead to inconsistent predictions, but if they are considered, then a prediction error for one label will be propagated to all related labels. Most of existing approaches either ignore label relations, or can learn only relations fitting a predefined imposed structure. This paper is motivated by the lack of a study analysing the compromise between handling label relations and limiting error propagation in GMLC, and by the fact that there is no known approach giving a control on that compromise to allow such a study. In this paper, a new meta-classifier with two main advantages is proposed for GMLC. Firstly, no predefined structure is imposed for learning label relations, and secondly, the meta-classifier is based on three measures giving control on the studied compromise. The studied compromise is analysed according to its impact on the classifier complexity and on hamming-loss evaluation measure. A comparison to three existing approaches shows that the proposed meta-classifier is competitive according to hammingloss evaluation measure, and it is the most stable classifier according to hamming-loss standard deviation.

show abstract

Rough Sets in Machine Learning: A Review

Bello

Falcón

2017

Studies in Computational Intelligence

View full text Add to dashboard Cite

Fuzzy Rough Decision Trees for Multi-label Classification

Cited by 4 publications

References 16 publications

New classification technique: fuzzy oblique decision tree

New classification technique: fuzzy oblique decision tree

Graded multi-label classification: Compromise between handling label relations and limiting error propagation

Rough Sets in Machine Learning: A Review

Contact Info

Product

Resources

About