Rough Set Based Decision Support—Models Easy to Interpret

Widz, Sebastian; Ślęzak, Dominik

doi:10.1007/978-1-4471-2760-4_6

Cited by 15 publications

(3 citation statements)

References 53 publications

(60 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We will also investigate feature selection and reduction methods, e.g. those based on rough sets [31,32]. We plan further work in the laboratory system, including the use of additional stressful substances and investigation of their impact on the behaviour of mussels.…”

Section: Future Workmentioning

confidence: 99%

Zebra mussels’ behaviour detection, extraction and classification using wavelets and kernel methods

Przymus

Rykaczewski

Winiewski

2014

Future Generation Computer Systems

View full text Add to dashboard Cite

Section: Future Workmentioning

confidence: 99%

Zebra mussels’ behaviour detection, extraction and classification using wavelets and kernel methods

Przymus

Rykaczewski

Winiewski

2014

Future Generation Computer Systems

View full text Add to dashboard Cite

“…Firstly, it is some kind of data compression which eases comprehension of analysed data. Secondly, most classifiers are better trained on non-redundant data (compare with [6]). Moreover, removal of superfluous attributes leads to smaller data, which will quicken classifier training.…”

Section: Introductionmentioning

confidence: 99%

Parallel Feature Selection Algorithm based on Rough Sets and Particle Swarm Optimization

Adamczyk¹

2014

Annals of Computer Science and Information Systems

View full text Add to dashboard Cite

Abstract-The aim of this paper is to propose a new method of solving feature selection problem. Foundations of presented algorithm lie in the theory of rough sets. Feature selection methods based on rough sets have been used with success in many data mining problems, but their weakness is their computational complexity. In order to overcome the above-mentioned problem, researches used diverse approximation techniques. This paper presents a new approach to approximation of reducts.Particle swarm optimization (PSO) is a stochastic metaheuristic similar to genetic algorithms. The idea is to see each potential solution as a particle with certain velocity flying through the problem space. The PSO finds optimal solutions by interactions of individuals in population. The main advantage of the PSO over genetic algorithms, is that PSO does not require complex operators such as crossover or mutation. It only uses simple mathematical operators to update position and velocity of each particle, which makes PSO computationally inexpensive in terms of both memory and runtime.The presented feature selection algorithm treats each feature subset as separate particle. Optimal subset, in terms of selected measure, is discovered as particles fly within the problem space. In order to speed up calculations and balance usage of hardware resources (processors, memory), parallel asynchronous version of PSO is applied. It is based on scheduling calculations of complex fitness function on slave processors, while the main one is responsible for updating particles data and checking algorithm's convergence. Applied approach scales well and provides balanced usage of given resources even if it is not feasible to use the same computational power of every processor, for instance when used resources are not homogeneous.Proposed method was tested on selected set of data sets from the UCI repository and results were compared to some of the classical algorithms.

show abstract

“…We introduce the concept of exception decision rules which are created during the reduct calculation process. The use of exceptions can be helpful in case we need to construct simplified decision models [10] and preserve more information about the original data set. We also explain the relationship of the proposed methods to the notion of decision bireducts and propose a new algorithm for decision bireducts calculation based on GMD function.…”

Section: Introductionmentioning

confidence: 99%

Generalized Majority Decision Reducts

Widz

Stawicki

2016

Annals of Computer Science and Information Systems

Self Cite

View full text Add to dashboard Cite

Abstract-We discuss several new methods for constructing approximate decision reducts from the rough set theory. We introduce generalized approximate majority decision reducts, which are an extension to standard approximate decision reducts known from literature but with improved calculation performance and complexity. We also discuss the relationship and differences of the new approximate decision reduct notion with so called decision bireducts -another type of approximate decision reducts.

show abstract

Rough Set Based Decision Support—Models Easy to Interpret

Cited by 15 publications

References 53 publications

Zebra mussels’ behaviour detection, extraction and classification using wavelets and kernel methods

Zebra mussels’ behaviour detection, extraction and classification using wavelets and kernel methods

Parallel Feature Selection Algorithm based on Rough Sets and Particle Swarm Optimization

Generalized Majority Decision Reducts

Contact Info

Product

Resources

About