A confidence-based late fusion framework for audio-visual biometric identification

Abstract

a b s t r a c tThis paper presents a confidence-based late fusion framework and its application to audio-visual biometric identification. We assign each biometric matcher a confidence value calculated from the matching scores it produces. Then a transformation of the matching scores is performed using a novel confidence-ratio (C-ratio) i.e., the ratio of a matcher confidence obtained at the test phase to the corresponding matcher confidence obtained at the training phase. We also propose modifications to the highest rank and Borda count rank fusion rules to incorporate the matcher confidence. We demonstrate by experiments that our proposed confidencebased fusion framework is more robust compared to the state-of-the-art late (score-and rank-level) fusion approaches.

Multimodal Data Fusion to Track Students’ Distress during Educational Gameplay

“…For multimodal data fusion, we implemented a decision-level fusion approach. Compared to feature-level fusion, which directly merges feature vectors, decision-level fusion integrates the predicted scores of each classifier (Alam et al, 2015) by categorizing results for each modality before fusing the outcome results. For the decision-level fusion, we combined the average confidence values of subdecisions into the final decision about the target outcome measure.…”

“…Here we computed each machine-learning classifier's probability scores for the state of distress. We then used confidence-based fusion (Alam et al, 2015) to create a classifier. We computed sub-decisions by confidence scores, using independent classifiers with different data sources.…”

Helicobacter Pylori infection detection from gastric X-ray images based on feature fusion and decision fusion

“…There have also been many studies on decision fusion. Much interest has been shown in confidence-based decision fusion schemes [28,29]. Alam et al [28] developed a fusion framework for audio-visual biometric identification.…”

“…Much interest has been shown in confidence-based decision fusion schemes [28,29]. Alam et al [28] developed a fusion framework for audio-visual biometric identification. The framework works well when input samples presented are contaminated by noise, e.g., detector noise, bit error and additive noise.…”