Generalization Bounds for Convolutional Neural Networks

Abstract

Convolutional neural networks (CNNs) have achieved breakthrough performances in a wide range of applications including image classification, semantic segmentation, and object detection. Previous research on characterizing the generalization ability of neural networks mostly focuses on fully connected neural networks (FNNs), regarding CNNs as a special case of FNNs without taking into account the special structure of convolutional layers. In this work, we propose a tighter generalization bound for CNNs by exploiting the sparse and permutation structure of its weight matrices. As the generalization bound relies on the spectral norm of weight matrices, we further study spectral norms of three commonly used convolution operations including standard convolution, depthwise convolution, and pointwise convolution. Theoretical and experimental results both demonstrate that our bounds for CNNs are tighter than existing bounds.

References in corpus (11)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• Understanding deep learning requires rethinking generalization
• The Loss Surfaces of Multilayer Networks
• On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima
• Loss Surfaces, Mode Connectivity, and Fast Ensembling of DNNs
• Matrix Completion has No Spurious Local Minimum
• Spectrally-normalized margin bounds for neural networks
• Stronger generalization bounds for deep nets via a compression approach
• An Information-Theoretic View for Deep Learning
• Improved Learning of One-hidden-layer Convolutional Neural Networks with Overlaps

Cited by in corpus (7)

• Recent advances in deep learning theory
• Statistical theory for image classification using deep convolutional neural networks with cross-entropy loss under the hierarchical max-pooling model
• CNNs Avoid Curse of Dimensionality by Learning on Patches
• Norm-based generalisation bounds for multi-class convolutional neural networks
• On the Robustness and Generalization of Deep Learning Driven Full Waveform Inversion
• Data-Adaptive Discriminative Feature Localization with Statistically Guaranteed Interpretation
• Non-asymptotic Excess Risk Bounds for Classification with Deep Convolutional Neural Networks