Arne Küderle

@felixkluge Any updates on this? :)

As soon as possible, latest end of June ;)

Great! I am a little concerned by the normalization. Shouldn't we scale the theoretical limits instead of the observed limits? Otherwise the performance index is not at all comparable between...

I think for the other values, we also need to "hard code" the value ranges to make them comparable. So we need to define what we consider best and worst,...

Do you think sigmoid is required? Or would "capping" be sufficient. Let's say for duration error, we define no error 0s -> 1, worst case 60s -> 0. If the...

> For duration error within [0,inf), f(x) = 1 - exp(-x) could make sense (as sigmoid would rather consider (-inf, inf)). Not sure whether either transformation or capping (with an...

@felixkluge Do you have time to look into this?

Notes: - We should keep the oldmodels around (unless newly trained are really identical -> unlikely) - The full retraining code should be kept around as a `no_exc` example ->...

Fixed in #186

Might still be useful for the future, but out of scope for now. The mobilise-D method is focused on strides and until some major additional research project investigates other output...