We present a novel language identification technique using our recently developed deep-structured conditional random fields (CRFs). The deep-structured CRF is a multi-layer CRF mo...
Supervised learning uses a training set of labeled examples to compute a classifier which is a mapping from feature vectors to class labels. The success of a learning algorithm i...
—The two-path algorithm is a well-known approach for overcoming the dead-lock problem in echo cancellation systems. Typically, a fixed foreground filter is producing the echo c...
A myriad of recent activities can be seen towards dynamic workflow composition for processing complex and data intensive problems. Meanwhile, the simultaneous emergence of the gri...
David Chiu, Sagar Deshpande, Gagan Agrawal, Rongxi...
Performance evaluation of complex systems is a critical issue and bounds computation provides confidence about service quality, reliability, etc. of such systems. The stochastic ...