Evaluating intelligent systems with performance uncertainty in large test spaces
Mark Thompson · 2010
This paper proposes a new methodology, the Spatial Variance Bounding Methodology, for designing tests/experiments for intelligent systems based on a meta-model method for characterizing a discontinuous, stochastic system. The intent of these designs is to provide a framework for evaluating the performance of these intelligent systems across large test spaces that may take months or years to complete via traditional test/experiment design methods.