Why explaining away concept is not applicable in restricted Boltzmann machines? Their hidden units form a V structure from which probabilistic influence can flow given the observed visible variable. Why is this a problem in deep belief nets?
It would be too lengthy to do the derivation here. Would highly recommend watching this video, where Hugo Larochelle has derived the conditional independence of hidden units in RBMs.
Hope this video answers all of your questions, which seem to spur from conditional independence of RBMs. If not, it would be helpful if you could elaborate your question and provide some more context.