Characterizing the Outlying Feature Set of Groups

Outlying feature set of groups is useful in many applicationscenarios.However, most of existing literatures focusedoncharacterizing outlying feature set of individuals rather than group level. A method that can identify outlying feature setof groups effectively from large scale dataset is not yet available. This paper aims to tackle this challenge by proposing a novelgroup outlying feature set identification algorithm, named GOFSI, which can identify the outlying feature set at the group level automatically. The Experiments on both synthetic and real-world data sets confirmed the effectiveness of ourmethod.