A Spline Tour of Deep Learning: The Scattering Way

Embark on a 28-minute journey exploring the fascinating world of deep learning with Richard Baraniuk from Rice University. Delve into the remarkable progress made in computational problems using deep neural networks trained on massive datasets. Examine the fundamental questions surrounding deep learning methods, including their effectiveness, applicability, and potential improvements. Gain insights into the implications of the current lack of understanding for consumers, practitioners, and researchers in machine learning. Discover recent advancements towards developing a theory of deep learning based on rigorous mathematical principles. Focus on the connection between deep networks and spline approximation, providing a geometric interpretation of data processing in deep networks. Explore the intriguing special case of the Scattering Network, offering a unique perspective on deep learning architectures.