Incremental updating approximations in probabilistic rough sets under the variation of attributes

The attribute set in an information system evolves in time when new information arrives. Both lower and upper approximations of a concept will change dynamically when attributes vary. Inspired by the former incremental algorithm in Pawlak rough sets, this paper focuses on new strategies of dynamically updating approximations in probabilistic rough sets and investigates four propositions of updating approximations under probabilistic rough sets. Two incremental algorithms based on adding attributes and deleting attributes under probabilistic rough sets are proposed, respectively. The experiments on five data sets from UCI and a genome data with thousand attributes validate the feasibility of the proposed incremental approaches.