Open Access

Fig. 9.

image

Download original image

Calibration tests of the density estimator on four different survey areas. Each contour represents an output of a newly retrained NDE with the same training parameters. The only difference is in the seeds used for generating random numbers, for example, initialization of weights in the neural network. The tests are done for the same testing XOD. We can see that the NDE is well-trained in that regard it produces consistent results and does not show large deviations only due to small differences in the initialization of its neural networks.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.