A little bit of hype can build excitement about potential, while too much leads to false hopes and misguided planning assumptions. "Right now, the myths surrounding artificial intelligence (AI) are rampant," says Alexander Linden, research vice president at Gartner. "Wisely for now, most organizations' commitments are tentative and more oriented toward experimenting and learning, rather than trying to transform their enterprise or industry as fast as they can." Enterprise architecture and technology innovation leaders must walk a fine line between embracing and overplaying AI technologies' role in delivering business value for digital business. "Leaders shouldn't trust any of the myths and hype around AI. Instead, they must become centers of expertise if they are going to educate senior business executives on the real benefits -- and shortcomings -- of AI," says Linden.

Restricted Boltzmann machines~(RBMs) and conditional RBMs~(CRBMs) are popular models for a wide range of applications. In previous work, learning on such models has been dominated by contrastive divergence~(CD) and its variants. Belief propagation~(BP) algorithms are believed to be slow for structured prediction on conditional RBMs~(e.g., Mnih et al. [2011]), and not as good as CD when applied in learning~(e.g., Larochelle et al. [2012]). In this work, we present a matrix-based implementation of belief propagation algorithms on CRBMs, which is easily scalable to tens of thousands of visible and hidden units. We demonstrate that, in both maximum likelihood and max-margin learning, training conditional RBMs with BP as the inference routine can provide significantly better results than current state-of-the-art CD methods on structured prediction problems. We also include practical guidelines on training CRBMs with BP, and some insights on the interaction of learning and inference algorithms for CRBMs.