Abstract--Rough face alignments lead to suboptimal performance of face identification systems. In this study, we present a novel approach for identifying genders from facial images...
—Discriminant feature extraction plays a central role in pattern recognition and classification. Linear Discriminant Analysis (LDA) is a traditional algorithm for supervised feat...
We propose a new loss function for discriminative learning of Markov random fields, which is an intermediate loss function between the sequential loss and the pointwise loss. We s...
Diagnostic tasks involve identifying faulty components from observations of symptomatic device behavior. This paper presents a general diagnostic theory that uses the perspective ...
In this paper, we propose a new semi-supervised training method for Gaussian Mixture Models. We add a conditional entropy minimizer to the maximum mutual information criteria, whi...