Maximum-likelihood dissimilarities in image recognition with deep neural networks

In this paper we focus on the image recognition problem in the case of a small sample size based on the nearest neighbor rule and matching high-dimensional feature vectors extracted with a deep convolutional neural network. We propose a novel recognition algorithm based on the maximum likelihood method for the joint density of dissimilarities between the observed image and available instances in a training set. This likelihood is estimated using the known asymptotically normally distribution of the Jensen-Shannon divergence between image features, if the latter can be treated as probability density estimates. This asymptotic behavior is in agreement with the well-known experimental estimates of the distributions of dissimilarity distances between the high-dimensional vectors. The experimental study in unconstrained face recognition for the LFW (Labeled Faces in the Wild) and YTF (YouTube Faces) datasets demonstrated that the proposed approach makes it possible to increase the recognition accuracy by 1-5% when compared with conventional classifiers.