iMTE

논문 제목 : Grad-CAM: Why did you say that? 논문 주소 : https://arxiv.org/abs/1611.07450 Grad-CAM: Why did you say that? We propose a technique for making Convolutional Neural Network (CNN)-based models more transparent by visualizing input regions that are 'important' for predictions -- or visual explanations. Our approach, called Gradient-weighted Class Activation Mapping arxiv.org 주요 내용 정리: 1) Grad-C..