Decision Fusion Using a Multi-linear Classiier

The contribution of this paper is twofold: (1) to formulate a fusion problem encountered in the design of a multi-modal identity veriication system as a particular classiication problem, (2) to propose a simple classiier to solve this problem. The multi-modal identity veriication system under consideration is built of d modalities in parallel, each one delivering as output a scalar number, called score, stating how much the claimed identity is veriied. A fusion module receiving as input the d scores has to take a binary decision: accept or reject identity. We have formulated the design of this fusion module as a particular classiication problem and developed a clas-siier based on a combination of half-spaces in the d-dimensional space to solve the problem. We call this classiier multi-linear classiier in reference to the use of several half-spaces, each one building a linear classiier. The performance of the fusion module has been evaluated on a multi-modal database.