Deep neural network architectures in geo-inverse problems  (Joint with IP seminar)  

 Abstract:  We will present recent advances in designing deep neural networks that shed light on inverse problems in seismology. We highlight both insights from analysis as well as learning data and coefficient or parameter representations. Joint research with R. Balestriero, J. Bruna, I. Dokmani\’{c}, M. Lassas, L. Seydoux and C. Wong