Applying Bayes based classifiers for decision fusion in a multi-modal identity verification system

The contribution of this paper is threefold: (1) to formulate a decision fusion problem encountered in the design of a multi-modal identity verification system as a particular classification problem, (2) to propose a general Bayes based classifier framework to solve this problem, (3) to derive and to compare two specific particularizations from this general framework. The multi-modal identity verification system under consideration is built of n modalities in parallel, each one delivering as output a scalar number, called score, stating how well the claimed identity is verified. A fusion module receiving as input the n scores has to take a binary decision: accept or reject identity. We have solved this fusion problem using Bayes based classifiers applied in two specific cases: one using Gaussian distributions and another one using distributions from the exponential family with equal dispersion parameters, which leads to the logistic regression model. The effect of the a priori probabilities is highlighted in this context. The performances of these two fusion modules have been evaluated and compared with other approaches on a multimodal database, containing both vocal and visual modali-