Improve Generalization and Robustness of Neural Networks via Weight Scale Shifting Invariant Regularizations

通过权重尺度偏移不变性正则化改善神经网络的泛化性和鲁棒性.

The Geometric Occam’s Razor Implicit in Deep Learning

深度学习中隐含的几何奥卡姆剃刀.

Implicit Gradient Regularization

隐式梯度正则化方法.

Spectral Norm Regularization for Improving the Generalizability of Deep Learning

使用谱正则化改进深度学习的泛化性.

Understanding the Role of Individual Units in a Deep Neural Network

理解卷积神经网络中单个卷积核的作用.

High Frequency Component Helps Explain the Generalization of Convolutional Neural Networks

从频率角度讨论卷积神经网络的泛化能力.