Lecture Title: Machines that learn: big data or explanatory models? Abstract: A leading question about machines that learn concerns two distinct styles of learning. Will they turn out to depend more on probabilistic models that explain the data, or on networks that react to data and are trained on data at ever greater scale? In machine vision systems, for instance, this boils down to the comparative roles of two paradigms: analysis-by-synthesis versus empirical recognisers. Each approach has its strengths, and empirical recognisers especially have made great strides in performance in the last few years, through deep learning.