Factorized Hidden Layer (FHL) adaptation has been proposed for speaker adaptation of deep neural network (DNN) based acoustic models. In FHL adaptation, a speaker-dependent (SD) transformation matrix and an SD bias are included in addition to the standard affine transformation. The SD transformation is a linear combination of rank-1 matrices whereas the SD bias is a linear combination of vector...

Rank transformations are frequently employed in numerical experiments involving a computational model, especially in the context of sensitivity and uncertainty analyses. Response surface replacement and parameter screening are tasks which may benefit from a rank transformation. Ranks can cope with nonlinear (albeit monotonic) input-output distributions, allowing the use of linear regression tec...

lorentz characterized the almost convergence through the concept of uniform convergence of de lavallée-poussin mean. in this paper, we generalize the notion of almost convergence by using the concept ofinvariant mean and the generalized de la vallée-poussin mean. we determine the bounded linear operators forthe generalized σ-conservative, σ-regular and σ-coercive matrices.

This paper considers the problem of 3D reconstruction from 2D points in one or more images and auxiliary information about the corresponding 3D features : alignments, coplanarities, ratios of lengths or symmetries are known. Our first contribution is a necessary and sufficient criterion that indicates whether a dataset, or subsets thereof, defines a rigid reconstruction up to scale and translat...