Improved acoustic modeling can significantly decrease the error rate in large-vocabulary speech recognition. Our approach to the problem is twofold. We first propose a scheme that...
The key problem to be faced when building a HMM-based continuous speech recogniser is maintaining the balance between model complexity and available training data. For large vocab...
Database Semantics (DBS) is a computational model of how communicating with natural language works. at a level of abstraction which may be applied to natural and artificial agents...
The use of net ecosystem exchange (NEE) data to estimate carbon transfer coefficients is investigated in the context of a deterministic compartmental carbon sequestration system. ...
: The main goal when synthesizing robust residual generators, for diagnosis and supervision, is to attenuate influence from model uncertainty on the residual while keeping fault de...