深度神经网络是否具有内在的奥卡姆剃刀原理？（英文版）

英文标题：Do deep neural networks have an inbuilt Occam's razor?中文摘要：研究表明，深度神经网络成功的关键在于结构化数据，以及拥有一种像奥卡姆剃刀一般的归纳偏差，使其能够对抗复杂度呈指数级增长的函数，通过引入贝叶斯图像和随机梯度下降等方法，我们试图将网络架构、训练算法和数据结构三个组件相互分离。英文摘要：The remarkable performance of overparameterized deep neural networks (DNNs)must arise from an interplay between network architecture, training algorithms,and structure in the data. To disentangle these three components, we apply aBayesian picture, based on the functions expressed by a DNN, to supervisedlearning. The pr