In Situ Adaptive Tabulation for Machine Learning скачать в хорошем качестве

In Situ Adaptive Tabulation for Machine Learning

ISAT is a nonlinear empirical modeling approach that adaptively builds models as new information is requested. ISAT has been applied to simulations of turbulent reacting flows as well as nonlinear model predictive control (NMPC) to make real-time applications feasible.  Like artificial neural networks, ISAT is a storage and retrieval method.  Unlike neural nets, ISAT is adaptive to extrapolation.  In addition, ISAT learns as it goes so there is no specified amount of training data before it starts performing. Source code is available from http://apm.byu.edu/prism/index.php/Pr...