Kernel methods have been successfully applied to many machine learning problems. Nevertheless, since the performance of kernel methods depends heavily on the type of kernels being...
Tianbao Yang, Mehrdad Mahdavi, Rong Jin, Jinfeng Y...
In many dynamic matching applications—especially high-stakes ones—the competitive ratios of prior-free online algorithms are unacceptably poor. The algorithm should take distr...
John P. Dickerson, Ariel D. Procaccia, Tuomas Sand...
—This paper describes a new technique, which automatically selects the most appropriate developers for fixing the fault represented by a failing test case, and provides a diagno...
Underlying biometrics are biological tissues that evolve over time. Hence, biometric authentication (and recognition in general) is a dynamic pattern recognition problem. We propos...
Norman Poh, Josef Kittler, Raymond S. Smith, Jose ...
We present a new approach based on Partial Differential Equations (PDE) to restore noisy blurred images. After studying the methods to denoise images, staying as close as possible...