چکیده:
Vast majority of data mining algorithms have been designed to work on centralized data, unfortunately however, almost all of nowadays data sets are distributed both geographically and conceptually. Due to privacy and computation cost, centralizing distributed data sets before analyzing them is undoubtedly impractical. In this paper, we present a framework for clustering distributed data which takes into account privacy and computation cost. To do that, we remove uncertain instances and just send the label of the other instances to the central location. To remove the uncertain instances, we develop a new instance weighting method based on fuzzy and rough set theory. The achieved results on well-known data verify effectiveness of the proposed method compared to previous works.
خلاصه ماشینی:
A Distributed Clustering Approach for Heterogeneous Environments Using Fuzzy Rough Set Theory Niloofar Mozafari Department of Designing & System Operation, Regional Information Center for Science and Technology, RICeST, Shiraz, IRAN Corresponding Author: mozafari Hricest.
Samatova, Ostrouchov, Geist & Melechko (2005) presented another hierarchical clustering in distributed environments that send a representative from each cluster to a central location.
There are also some methods for distributed clustering in homogeneous data that work well in distributed environments but they do not specifically address the privacy issues (Tasoulis & Vrahatis, 2004), (Dhillon & Modha, 2002).
A density based clustering in distributed environments was proposed in (Santos, Syed, Naldi, Campello & Sander, 2019).
For selecting the appropriate labels, we propose a new instance weighting method based on fuzzy and rough set theory.
Our Fuzzy Rough set Instance Weighting (FRIW) gives weight to each instance and instead of sending the entire data, the label of instances with higher weights are just sent to the central location.
Instead of sending the entire instances with all of their features to central location, the labels of selected data are just sent.
As it is obvious, in this data set outlier instances which are far from the core (central) of the cluster, have the minimum weights.
For example, in Pendig data set; with removing the boundary instances; the number of label of instances in the central location decreases from 44964 to 25526.
In order to select the appropriate labels, we propose a new instance weighting method based on fuzzy and rough set theory.