Closed benwulfe closed 1 year ago
probably not viable because you need 6 points for each sample in the training set (mean/var for each element) Let's leave open with the scope of looking at what input data is available for training to make this viable.
Sprint 5
Ruben is concerned that this might not be feasible but will check at the Friday work session.
If elements have some amount of covariance, it may be useful to train a single model to produce all means and variances. This may help us in situations where we have ground truth for some elements, but not others - possibly giving us a better means/variance for elements especially in cases of limited data.