Conditions for learning in generalized tandem networks | IEEE Conference Publication | IEEE Xplore