The use of the statistical technique of mixture model analysis as a tool for early prediction of fault-prone program modules is investigated. The Expectation-Maximum likelihood (E...
Modeling of individual users is a promising way of improving the performance of spoken dialogue systems deployed for the general public and utilized repeatedly. We define "im...
We analyze the effect of errors in branch predictors, a representative example of speculative processor subsystems, to motivate the necessity for fault tolerance in such subsystem...
This paper proposes and evaluates several alternate design choices for common prediction metrics employed by neighborhood-based collaborative filtering approach. It first explores ...
The basic idea presented in this paper is to use interference aware routing with predicted interference. Noise and other values are measured over a long time and within a wide are...