Are there any pathological problems for which deep belief networks fail catastrophically?
> en.wikipedia.org/wiki/Dee
To begin with, I’d like to dissect this question…. Superficially, it seems to mix apples and oranges. “Pathological problems” are generally mental/medical in nature, and “deep belief networks” are basically deep neural networks. It would seem this question applies to psychological applications of deep neural networks. IMHO, I don’t think there has been enough work simulating mental states, much less pathological mental states, with deep neural networks to determine whether or not they fail given pathological problems. Please correct me if I’m wrong.