增采样
人工智能
联营
混叠
计算机科学
Sinc函数
卷积神经网络
模式识别(心理学)
稳健性(进化)
信号处理
算法
图像处理
卷积(计算机科学)
计算机视觉
棱锥(几何)
抽取
背景(考古学)
推论
傅里叶变换
频域
小波
多维信号处理
信号重构
工件(错误)
采样(信号处理)
消除混叠
图像(数学)
迭代重建
核(代数)
数学
奈奎斯特-香农抽样定理
人工神经网络
深度学习
双正交系统
剪切波
对抗制
作者
Julia Grabinski,Steffen Jung,Janis Keuper,Margret Keuper
标识
DOI:10.1007/s11263-026-03005-9
摘要
Abstract Convolutional Neural Networks (CNNs) are successful in various computer vision tasks. From an image and signal processing point of view, this success is counter-intuitive, as the inherent spatial pyramid design of most CNNs is apparently violating basic signal processing laws, i.e. the Sampling Theorem in their downsampling operations. This issue has been broadly neglected until recent work in the context of adversarial attacks and distribution shifts showed that there is a strong correlation between the vulnerability of CNNs and aliasing artifacts induced by bandlimit-violating downsampling. As a remedy, we propose an alias-free downsampling operation in the frequency domain, denoted Frequency Low Cut Pooling (FLC Pooling) which we further extend to Aliasing and Sinc Artifact-free Pooling (ASAP) . ASAP is alias-free and removes further artifacts from sinc-interpolation. Our experimental evaluation on ImageNet-1k, ImageNet-C and CIFAR datasets on various CNN architectures demonstrates that networks using FLC Pooling and ASAP as downsampling methods learn more stable features as measured by their robustness against common corruptions and adversarial attacks, while maintaining a clean accuracy similar to the respective baseline models.
科研通智能强力驱动
Strongly Powered by AbleSci AI