One-class classification with application to forensic analysis

The analysis of broken glass is forensically important to reconstruct the events of a criminal act. In particular, the comparison between the glass fragments found on a suspect (recovered cases) and those collected at the crime scene (control cases) may help the police to identify the offender(s) correctly. The forensic issue can be framed as a one-class classification problem. One-class classification is a recently emerging and special classification task, where only one class is fully known (the so-calledtargetclass), whereas information on the others is completely missing. We propose to consider Gini's classicaltransvariation probabilityas a measure of typicality, i.e. a measure of resemblance between an observation and a set of well-known objects (the control cases). The aim of the proposedtransvariation-based one-class classifieris to identify the best boundary around the target class, i.e. to recognize as many target objects as possible while rejecting all those deviating from this class.