The ANN is trained on random values from an hypercube specified by
boundary::MinMaxBoundaryValues{Float64}A test in the C onnxWrapper library to check if the trained ANN is evaluated inside or outside of this hyper-cube is useful to prevent evaluations outside of the trained area. Or at least notify the user / simulation environment that the result should not be trusted.
The ANN is trained on random values from an hypercube specified by
A test in the C onnxWrapper library to check if the trained ANN is evaluated inside or outside of this hyper-cube is useful to prevent evaluations outside of the trained area. Or at least notify the user / simulation environment that the result should not be trusted.