Biometric score fusion through discriminative training

In the multibiometric systems, various matcher/modality scores are fused together to provide better performance than the individual matcher scores. In [1] the authors have proposed a likelihood ratio test (LRT) based fusion technique for the biometric verification task that outperformed several other classifiers. They model the genuine and the imposter densities by the finite Gaussian mixture models (GMM, a generative model) whose parameters are estimated using the maximum likelihood (ML) criteria. Lately, the discriminative training methods and models have been shown to provide additional accuracy gains over the generative models, in multiple applications such as the speech recognition, verification and text analytics[5, 7]. These gains are based on the fact that the discriminative models are able to partially compensate for the unavoidable mismatch, which is always present between the specified statistical model (GMM in this case) and the true distribution of the data which is unknown. In this paper, we propose to use a discriminative method to estimate the GMM density parameters using the maximum accept and reject (MARS) criteria[8]. The test results using the proposed method on the NIST-BSSRI multimodal dataset indicate improved verification performance over a very competitive maximum likelihood (ML) trained system proposed in [1].