Leveraging Sparse Linear Layers for Debuggable Deep Networks
Leveraging Sparse Linear Layers for Debuggable Deep Networks
复制标题
DOI：
--
复制
发表时间：
2021-05
期刊：
影响因子：
--
通讯作者：
Eric Wong;Shibani Santurkar;A. Madry
中科院分区：
文献类型：
--
作者：
Eric Wong;Shibani Santurkar;A. Madry
我们展示了在学习到的深度特征表示上拟合稀疏线性模型如何能够产生更易于调试的神经网络。这些网络保持了很高的准确性，同时也更易于人类解释，正如我们通过数值和人工实验定量证明的那样。我们进一步说明了由此产生的稀疏解释如何有助于识别虚假相关性、解释错误分类以及诊断视觉和语言任务中的模型偏差。我们工具包的代码可在https://github.com/madrylab/debuggabledeepnetworks找到。
We show how fitting sparse linear models over learned deep feature representations can lead to more debuggable neural networks. These networks remain highly accurate while also being more amenable to human interpretation, as we demonstrate quantiatively via numerical and human experiments. We further illustrate how the resulting sparse explanations can help to identify spurious correlations, explain misclassifications, and diagnose model biases in vision and language tasks. The code for our toolkit can be found at https://github.com/madrylab/debuggabledeepnetworks.