Dynamical Isometry and a Mean Field Theory of RNNs: Gating Enables Signal Propagation in Recurrent Neural Networks

Minmin Chen, Jeffrey Pennington, Samuel S. Schoenholz. Dynamical Isometry and a Mean Field Theory of RNNs: Gating Enables Signal Propagation in Recurrent Neural Networks. In Jennifer G. Dy, Andreas Krause 0001, editors, Proceedings of the 35th International Conference on Machine Learning, ICML 2018, Stockholmsmässan, Stockholm, Sweden, July 10-15, 2018. Volume 80 of JMLR Workshop and Conference Proceedings, pages 872-881, JMLR.org, 2018. [doi]

Abstract

Abstract is missing.