Institute of Information Theory and Automation

Limitations of Shallow Neural Networks

Lecturer: Dr. Věra Kůrková
Institute: Institute of Computer Science, Czech Academy of Sciences
Date and time: 10.03.2017 - 10:00
Room: 25
Department: Image Processing (ZOI)


Recent successes of deep networks pose a theoretical question: When are deep nets provably better than the shallow ones? Using probabilistic and geometric properties of high-dimensional spaces we will show that for most common types of computational units, almost any uniformly randomly chosen function on a sufficiently large domain cannot be computed by a reasonably sparse shallow network. We will also discuss connections with the No Free Lunch Theorem, the central paradox of coding theory, and pseudo-noise sequences.
Institute of Information Theory and Automation