Graph interpolating activation improves both natural and robust accuracies in data-efficient deep learning

DOI10.1017/S0956792520000406zbMath1505.68039arXiv1907.06800OpenAlexW3117053485MaRDI QIDQ5014842

Bao Wang, Stanley J. Osher

Publication date: 8 December 2021

Published in: European Journal of Applied Mathematics (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/1907.06800

zbMATH Keywords

robustnessadversarial defensedata-dependent activationdata-efficient learningmanifold-learningsample-efficiency

Mathematics Subject Classification ID

Artificial neural networks and deep learning (68T07)

Related Items (3)

EnResNet: ResNets Ensemble via the Feynman--Kac Formalism for Adversarial Defense and Beyond ⋮ How does momentum benefit deep neural networks architecture design? A few case studies ⋮ Connections between deep learning and partial differential equations

Uses Software

• darch
• MNIST
• FLANN
• CIFAR
• PyTorch
• ImageNet
• Adam
• GitHub
• AlexNet
• LaplacianSmoothing-GradientDescent

Cites Work

• Unnamed Item
• Adversarial defense via the data-dependent activation, total variation minimization, and adversarial training
• A Fast Learning Algorithm for Deep Belief Nets

This page was built for publication: Graph interpolating activation improves both natural and robust accuracies in data-efficient deep learning