On Smoother Attributions using Neural Stochastic Differential Equations
On Smoother Attributions using Neural Stochastic Differential Equations
复制标题
DOI：
10.24963/ijcai.2021/73
复制
发表时间：
2021-08
期刊：
影响因子：
--
通讯作者：
Sumit Kumar Jha;Rickard Ewetz;Alvaro Velasquez;Susmit Jha
中科院分区：
文献类型：
--
作者：
Sumit Kumar Jha;Rickard Ewetz;Alvaro Velasquez;Susmit Jha
最近已经开发了几种方法来计算神经网络对输入特征的预测的属性。然而，这些现有的方法计算属性是嘈杂的，不鲁棒的输入的小扰动。本文使用最近确定的动力系统和残差神经网络之间的连接，表明神经随机微分方程（SDES）计算的属性是噪音较小，视觉上更清晰，定量上更强大。利用动力系统理论，我们从理论上分析了这些属性的鲁棒性。我们还通过使用ResNet-50，WideResNet-101模型和ResNeXt-101模型计算ImageNet图像的属性，实验证明了我们的方法在提供更平滑，视觉更清晰和定量鲁棒属性方面的有效性。
Several methods have recently been developed for computing attributions of a neural network's prediction over the input features. However, these existing approaches for computing attributions are noisy and not robust to small perturbations of the input. This paper uses the recently identified connection between dynamical systems and residual neural networks to show that the attributions computed over neural stochastic differential equations (SDEs) are less noisy, visually sharper, and quantitatively more robust. Using dynamical systems theory, we theoretically analyze the robustness of these attributions. We also experimentally demonstrate the efficacy of our approach in providing smoother, visually sharper and quantitatively robust attributions by computing attributions for ImageNet images using ResNet-50, WideResNet-101 models and ResNeXt-101 models.