A fuzzy rough set based fitting approach for fuzzy set-valued information system
- PDF / 637,044 Bytes
- 10 Pages / 595.276 x 790.866 pts Page_size
- 29 Downloads / 175 Views
ORIGINAL RESEARCH
A fuzzy rough set based fitting approach for fuzzy set-valued information system Waseem Ahmed1
•
M. M. Sufyan Beg2 • Tanvir Ahmad1
Received: 18 May 2018 / Accepted: 2 April 2019 Bharati Vidyapeeth’s Institute of Computer Applications and Management 2019
Abstract Information systems comprising of multiple attribute values are known as set-valued information systems (SIS). This paper proposes a novel fuzzy set-valued information system (FSIS) as a generalized model of SIS. SIS is known as FSIS if some of its set-values are fuzzy set-values. Existing fuzzy rough set (FRS) model for SIS does not fit a dataset well and can lead to samples misclassification since sometimes there exist a large overlap between samples, so a fitting FRS model is introduced for handling FSIS. Fuzzy similarity relations obtained from each fuzzy set-valued attribute are parameterized using two parameter values to characterize the fuzzy information granules and then based on it, fuzzy lower and upper approximations are reconstructed. This approach can fit FSIS well and prevent samples misclassification. Later, the relative reduct of FSIS is computed using greedy forward algorithm. The effectiveness of proposed FSIS using fitting FRS model approach is proved by conducting a comparative study of proposed FSIS with existing information systems.
& Waseem Ahmed [email protected] M. M. Sufyan Beg [email protected] Tanvir Ahmad [email protected] 1
Department of Computer Engineering, Jamia Millia Islamia, New Delhi 110025, India
2
Department of Computer Engineering, Aligarh Muslim University, Aligarh, UP 202001, India
Keywords Attributes reduct Fuzzy similarity relation Fuzzy set-valued information system Dependency function
1 Introduction Rough set theory (RST) has been gaining a lot of attention from research community due to its usage in the analysis of vague and imperfect data [1–13]. The essential usage of RST is to evaluate the attribute reduct in an IS [1, 7]. Conventional rough set were inefficient in handling continuous or real-valued IS, so to tackle this issue, fuzzy rough set (FRS) is being widely used [14, 15]. In FRS, similarity relation is used in place of equivalence relation for handling real-valued attributes and this similarity degree between samples, measures the extent to which they are similar [7, 16]. Initially, RST considered the case of single valued attributes [1, 11, 17], but in reality, multiple values can exist for an attribute, that results in the construction of SIS [2, 18, 19]. Orlowska [18, 19] introduces the concept of non-deterministic information systems and investigated SIS. Lipski [20] examined SIS in terms of incomplete information systems, which was further studied by Kryszkiewicz [21] and Grzymala-Busse [22, 23] in more detail. Guan and Wang [2] considered SIS as a generalization of single valued IS. Later, FRS model for SIS was investigated by Dai and Tian [24] and crisp tolerance relation was replaced by fuzzy tolerance relation. Attributes having fuzzy set va
Data Loading...
