I don't understand how people can defend his detractors in this particular case. Are you telling me that an image upsampling model that does not contain hard coded bias, and trained on unbiased data will produced biased result? Especially the kind of biased result represented by the error made by the original tweeter who fucked up?
Just curious, but what "error" did the original tweeter make? Did anyone really expect the model to accurately reconstruct the original photo starting from a pixelated mess? That makes no sense to anyone with even a passing knowledge of ML. You're always going to get craploads of bias and variance (i.e. blatant inaccuracy, over and above the bias) in such a setting, even starting from "ideal, unbiased" data. The problem domain is at issue here.
Yeah I get your point. But I guess for this model you can kinda have a concept of the "ideal" training set, where all high frequency features appear at the same rate as in real world.
