Improving Robustness Without Sacrificing Accuracy with Patch Gaussian\n Augmentation

2019/06/06 by Raphael Gontijo Lopes, Lopes, Raphael Gontijo, Dong Yin +7 · 12 citations
Computer Science · #Advanced Neural Network Applications #Adversarial Robustness in Machine Learning #Anomaly Detection Techniques and Applications #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · pdf · doi:10.48550/arxiv.1906.02611

openalex publication_date 2019/06/06 · openalex created_date 2022/07/29 · openalex updated_date 2026/07/28

Abstract

Deploying machine learning systems in the real world requires both high\naccuracy on clean data and robustness to naturally occurring corruptions. While\narchitectural advances have led to improved accuracy, building robust models\nremains challenging. Prior work has argued that there is an inherent trade-off\nbetween robustness and accuracy, which is exemplified by standard data augment\ntechniques such as Cutout, which improves clean accuracy but not robustness,\nand additive Gaussian noise, which improves robustness but hurts accuracy. To\novercome this trade-off, we introduce Patch Gaussian, a simple augmentation\nscheme that adds noise to randomly selected patches in an input image. Models\ntrained with Patch Gaussian achieve state of the art on the CIFAR-10 and\nImageNetCommon Corruptions benchmarks while also improving accuracy on clean\ndata. We find that this augmentation leads to reduced sensitivity to high\nfrequency noise(similar to Gaussian) while retaining the ability to take\nadvantage of relevant high frequency information in the image (similar to\nCutout). Finally, we show that Patch Gaussian can be used in conjunction with\nother regularization methods and data augmentation policies such as\nAutoAugment, and improves performance on the COCO object detection benchmark.\n

Citations

• ImageNet classification with deep convolutional neural networks
• Random Erasing Data Augmentation
• DropBlock: A regularization method for convolutional networks
• Learning Multiple Layers of Features from Tiny Images

Cited by

• Stylized Synthetic Augmentation further improves Corruption Robustness
• Self-Training With Noisy Student Improves ImageNet Classification
• Meta Adversarial Training against Universal Patches
• WAInjectBench: Benchmarking Prompt Injection Detections for Web Agents
• PDA: Progressive Data Augmentation for General Robustness of Deep Neural Networks
• Learning Augmentation Distributions using Transformed Risk Minimization
• Viewmaker Networks: Learning Views for Unsupervised Representation Learning
• RoMA: Robust Model Adaptation for Offline Model-based Optimization
• Test-Time Adaptation to Distribution Shift by Confidence Maximization and Input Transformation
• DeeperForensics Challenge 2020 on Real-World Face Forgery Detection: Methods and Results
• Defending Against Image Corruptions Through Adversarial Augmentations
• Beyond topography: Topographic regularization improves robustness and reshapes representations in convolutional neural networks

Related

• A Fourier Perspective on Model Robustness in Computer Vision
• Improving robustness against common corruptions with frequency biased models
• Improving robustness against common corruptions by covariate shift adaptation
• A simple way to make neural networks robust against diverse image corruptions
• On Interaction Between Augmentations and Corruptions in Natural Corruption Robustness