A new class of kernels for object recognition based on local image feature representations are introduced in this paper. These kernels satisfy the Mercer condition and incorporate...
Invariant representations in object recognition systems are generally obtained by pooling feature vectors over spatially local neighborhoods. But pooling is not local in the featu...
Y-Lan Boureau, Nicolas Le Roux, francis bach, Jean...
Abstract. In this paper, we propose a novel representation, called Multiscale Block Local Binary Pattern (MB-LBP), and apply it to face recognition. The Local Binary Pattern (LBP) ...
Abstract--We present a new tensor-based morphometric framework that quantifies cortical shape variations using a local area element. The local area element is computed from the Rie...
We introduce and apply a genetic representation for analog electronic circuits based on the association of character strings extracted from the genome with the terminals and param...