The recognition of text in everyday scenes is made difficult by viewing conditions, unusual fonts, and lack of linguistic context. Most methods integrate a priori appearance info...
David Smith, Jacqueline Feild, Eric Learned-Miller
Analyzing data on-board a spacecraft as it is collected enables several advanced spacecraft capabilities, such as prioritizing observations to make the best use of limited bandwid...
Probabilistic language models are critical to applications in natural language processing that include speech recognition, optical character recognition, and interfaces for text e...
— Using the 1998 DARPA BSM data set collected at MIT’s Lincoln Labs to study intrusion detection systems, the performance of robust support vector machines (RVSMs) was compared...
Quadratic program relaxations are proposed as an alternative to linear program relaxations and tree reweighted belief propagation for the metric labeling or MAP estimation problem...