Fitting, Sensing and Learning

We are more than what we publish in research papers. We have insights, not just results and proofs. Luckily, there are ways to bring out these aspects, say through surveys and invited talks. I have been struggling to find the connection between sparse approximation, compressed sensing and learning problems: one senses that there are similarities and that if one learns to look at them in a certain light, they would fit in a puzzle together. Ron DeVore has a beautiful, mathematical exposition of the data fitting, sensing and learning problems worth reading.