Vytautas Šaltenis

@rq@river.group.lt lietuviška wikipedija išvis yra blogai. Trinti ir pradėti rašyti iš naujo.

@ly2en while we shouldn't think the layers are learning those specific facial features, I'd say it's a decent illustration for what they try to illustrate: the later the layer, the more semantic are the features that it optimises for.

@ly2en don't know, I never studied NNs down to the mathy details. But that's how I was explained these things by people in the know. And it follows intuitively: since your output layer has to know "the answer", and the only thing it talks to is the n-1st layer, the latter has to have "almost the answer", i.e. a representation that's semantically closest to the answer. Not a proof, of course, just a helpful intuition.