Deep learning(DL)had its coming out to the general pub-lic with the New York Times expose of Google's“secretive X laboratory”in 20121),and to the technology community with the huge leap in performance for speech recognition,natural language translation and image processing about the same time[1].The review by Bengio,Courville and Vin-cent[2]provides perspective,trying to interpret the dramatic successes seen,but most interesting for our discussion is their demonstrated mastery of a broad range machine learn-ing(ML)and artificial intelligence(AI)theories:Bayesian modelling,probabilistic graphical models,Markov models,causality,reinforcement learning,and so forth.While its well known to older machine learning researchers that the founders of deep learning are extremely knowledgable in the broader area,a popular view of deep learning sometimes seen in the media posits a different understanding:deep learning“works like the brain”and uses architectural recipes com-piled on systems such as Tensorflow,with little reference to statistical machine learning(SML).This article reviews the contributions of SML and considers their place in the broader ML landscape of the DL era.