作者: Sebastian Goldt , Marc Mézard , Florent Krzakala , Lenka Zdeborová
DOI: 10.1103/PHYSREVX.10.041044
关键词: Artificial intelligence 、 Data structure 、 Manifold (fluid mechanics) 、 Deep neural networks 、 Artificial neural network 、 Computer science
摘要: Understanding the reasons for the success of deep neural networks trained using stochastic gradient-based methods is a key open problem for the nascent theory of deep learning. The …