In this section, we evaluate our modeling approach on the 3D U-net case study to generate an analytical peak memory consumption model. In Experimental setup, the experimental setup, data collection, and sampling processing employed to generate the training and test data are described. In Model validation, the evaluation of the model generated by Pandora is presented, comparing the result against the validation data. Finally in Comparative evaluation with ML regression methods, we compare our modeling approach with other machine-learning based approaches that have proven to be promising candidates for performance modeling [16].
