Abstract:
An artificial intelligence testing apparatus may include processing circuitry configured to execute instructions that, when executed, cause the apparatus to create initial sample points based on a simulation received at the apparatus, and employ cyclic evaluation of the simulation until a stopping criteria is met. Employing the cyclic evaluation includes running the simulation at design points for a set of queries associated with a current iteration of the cyclic evaluation, training a set of meta-models of parameter space associated with the simulation for the current iteration, computing a set of metrics for the current iteration, and employing a selected sampling approach to select a new set of design points for a next iteration of the cyclic evaluation.