Comparison again re-implemented method.

DANMF [8], a network structure of alternating layers for matrix factorization where “factorizations of the basis and coding matrices are alternated along layers”, was re-implemented in the chosen framework. The loss was, at best, an order of 102–104 times worse than the average performance achieved by models implemented in this paper, for both input sizes. This potentially suggests an issue in the re-implementation or application to the simulated plasma data used. (TXT)