An Improved ACO Based Plug-in to Enhance the Interpretability of Fuzzy Rule Bases with Exceptions

Carmona, Pablo; Castro, Juan Luis

doi:10.1007/978-3-540-87527-7_2

Cited by 1 publication

(2 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our case, an enhanced ant colony‐based model elaborated by Carmona [25] is used to build the fuzzy rules. Their algorithm extends the syntax of the rules by associating more than one label to each input variable in the antecedent of the rule and by using other relational operators different from the usual ‘equal‐to’ operator.…”

Section: Proposed Systemmentioning

confidence: 99%

“…To apply ant colony meta‐heuristic method to the fuzzy rules building problem, the following steps have to be performed [24, 25]: (i) obtain a problem representation as a graph covered by ants; (ii) assign a heuristic preference to each ant's choice to generate the solution; (iii) initialise the pheromone trails; (iv) define a fitness function to be optimised; and finally (v) apply the ant colony algorithm to the problem. The complete set of compounded fuzzy rules after applying the utilised rules building model is given as follows

\begin{matrix} right leftthickmathspace.5em \\ R^{1} : boldI boldF A > falsefalse{low falsefalse}, T = falsefalse{medium falsefalse}, N_{D R} \div falsefalse{low, high} \\ m = falsefalse{low falsefalse}, D = falsefalse{low} boldT boldh bolde boldn Defect normal_Type = Warp \end{matrix}

\begin{matrix} right leftthickmathspace.5em \\ R^{2} : boldI boldF A \geq falsefalse{medium falsefalse}, T \geq falsefalse{medium falsefalse}, N_{D R} \div falsefalse{low, high} \\ m = falsefalse{high falsefalse}, D = falsefalse{low} boldT boldh bolde boldn Defect normal_Type = Weft \end{matrix}

\begin{matrix} right leftthickmathspace.5em \\ R^{3} : boldI boldF A \div falsefalse{low, medium falsefalse}, \end{matrix}

…”

Section: Proposed Systemmentioning

confidence: 99%

See 1 more Smart Citation

Soft computing applied to the build of textile defects inspection system

Darwish

2013

IET Computer Vision

View full text Add to dashboard Cite

The inspection of textile defects is challenging because of the large number of defects categories that are characterised by their imprecision and uncertainty. In this study, novel interval type-2 fuzzy system is proposed for resolving defects recognition problem of textile industries. The proposed system mixes interval type-2 fuzzy reasoning and swarm optimisation algorithm together in order to enhance the defects classification capabilities. Interval type-2 fuzzy logic is powerful in handling high level of indecisions in the human decision making process, including uncertainties in measurements of textile features and data used to calibrate the examination's parameters. Swarm intelligence algorithm is used to optimise parameters of the membership functions to increase the accuracy of fuzzy controller. Besides, the problem of fuzzy linguistic rules learning has been tackled by utilising ant colony meta-heuristic method to reduce the complexity of the inspection system. Excellent recogniser results on real textile samples, using this system, are demonstrated.

show abstract

Section: Proposed Systemmentioning

confidence: 99%