BriefGPT.xyz
Jan, 2021
形状还是纹理:理解卷积神经网络中的辨别特征
Shape or Texture: Understanding Discriminative Features in CNNs
HTML
PDF
Md Amirul Islam, Matthew Kowal, Patrick Esser, Sen Jia, Bjorn Ommer...
TL;DR
本研究通过设计一系列实验,发现卷积神经网络中大部分的形状信息在训练的前几个阶段就已经学习,这些信息主要由CNN的最后几层进行编码。此外,本文还表明形状的编码并不意味着编码的是局部像素语义信息。
Abstract
Contrasting the previous evidence that neurons in the later layers of a
convolutional neural network
(CNN) respond to complex object shapes, recent studies have shown that CNNs actually exhibit a `
texture bias
':
→