Explaining Failure: Investigation of Surprise and Expectation in CNNs
Explaining Failure: Investigation of Surprise and Expectation in CNNs
复制标题
DOI：
10.1109/cvprw50498.2020.00014
复制
发表时间：
2020-06
期刊：
影响因子：
--
通讯作者：
Thomas Hartley;K. Sidorov;C. Willis;David Marshall
中科院分区：
文献类型：
--
作者：
Thomas Hartley;K. Sidorov;C. Willis;David Marshall
人工智能卷积神经网络（CNN）已经扩展到日常使用中，需要更严格的方法来解释其内部工作原理。当前流行的技术（例如显着图）展示了网络如何通过根据像素的重要性对像素进行评分来在简单的级别上解释输入图像。在本文中，我们引入了惊喜和期望的概念，作为探索和可视化网络如何通过理解过滤器激活来学习对训练数据进行建模的手段。我们证明，这是一种强大的技术，可以帮助您了解网络对未见过的图像（与训练数据相比）的反应。我们还表明，我们的技术提供的见解使我们能够“修复”错误分类。我们的技术几乎可以用于所有类型的 CNN。我们使用 ImageNet 定性和定量地评估我们的方法。
Ai Convolutional Neural Networks (CNNs) have expanded into everyday use, more rigorous methods of explaining their inner workings are required. Current popular techniques, such as saliency maps, show how a network interprets an input image at a simple level by scoring pixels according to their importance. In this paper, we introduce the concept of surprise and expectation as means for exploring and visualising how a network learns to model the training data through the understanding of filter activations. We show that this is a powerful technique for understanding how the network reacts to an unseen image compared to the training data. We also show that the insights provided by our technique allows us to "fix" misclassifica- tions. Our technique can be used with nearly all types of CNN. We evaluate our method both qualitatively and quantitatively using ImageNet.