A Rough Sets Approach for Personalized Support of Face Recognition

Abstract

The activity of facial recognition is routine for most people; yet describing the process of recognition, or describing a face to be recognized reveals a great deal of complexity inherent in the activity. Eyewitness identification remains an important element in judicial proceedings. It is very convincing, yet it is not very accurate. We studied how people sorted a collection of facial photographs and found that individuals may have different strategies for similarity recognition. In our analysis of the data, we have identified two possible strategies. We apply rough set based attribute reduction methodology to this data in order to develop a test to identify which of these strategies an individual is likely to prefer. We hypothesize that by providing a personalized search and filter environment, individuals would be more adequately equipped to handle the complexity of the task, thereby increasing the accuracy of identifications. Furthermore, the rough set based analysis may help to more clearly identify the different strategies that individuals use for this task. This paper provides a description of the preliminary study, our computational approach that includes an important pre-processing step, discusses results from our evaluation, and provides a list of opportunities for future work.