文件列表:
评估隐式可解释性的后解释性【英文版】.pdf |
下载文档 |
资源简介
>
英文标题:Evaluating Post-hoc Interpretability with Intrinsic Interpretability中文摘要:本文对基于切片图像的乳腺癌转移检测问题,将时下最流行的用于深度神经网络(DNN)可解释性的后置解释方法(Post-hoc)和内置解释方法(Intrinsic)应用于一种新的可解释性 DNN,ProtoPNet,并比较了这两种方法的兴趣图像(Attribution maps)的相似性。结果显示,除 SmoothGrad 和 Occlusion 以外,其他后置方法兴趣图像的重合度很低。英文摘要:Despite Convolutional Neural Networks having reached human-level performancein some medical tasks, their clinical use has been hindered by their lack ofinterpretability. Two major interpretability strategies have been propose
加载中...
本文档仅能预览20页