Since it is extremely expensive to collect a large volume of handwriting samples, synthesized data are often used to enlarge the training set. We argue that, in order to generate ...
Nonlinear dynamical and control systems are an important source of applications for theories of computation over the the real numbers, since these systems are usually to complicate...
We describe a simple random-sampling based procedure for producing sparse matrix approximations. Our procedure and analysis are extremely simple: the analysis uses nothing more th...
We describe a methodology for retrieving document images from large extremely diverse collections. First we perform content extraction, that is the location and measurement of reg...
We describe a framework for decomposing the distortion between two images into a linear combination of components. Unlike conventional linear bases such as those in Fourier or wav...