Improving Shift Invariance in Convolutional Neural Networks with Translation Invariant Polyphase Sampling

arXiv:2404.07410v1 Announce Type: cross
Abstract: Downsampling operators break the shift invariance of convolutional neural networks (CNNs) and this affects the robustness of features learned by CNNs when dealing with even small pixel-level shift. Through a large-scale correlation analysis framework, we study shift invariance of CNNs by inspecting existing downsampling operators in terms of their maximum-sampling bias (MSB), and find that MSB is negatively correlated with shift invariance. Based on this crucial insight, we propose a learnable pooling operator called Translation …

abstract analysis arxiv cnns convolutional neural networks correlation cs.cv cs.lg downsampling features framework improving networks neural networks operators pixel robustness sampling scale shift small study through translation type