Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps
By Karen Simonyan, Andrea Vedaldi, Andrew Zisserman. 2013
8,399 citations recorded by the source corpus.
This paper addresses the visualisation of image classification models, learnt using deep Convolutional Networks (ConvNets). We consider two visualisation techniques, based on computing the gradient of the class score with respect to the input image. The first one generates an image, which maximises the class score [Erhan et al., 2009], thus visualising the notion of the class, captured by a ConvNet. The second technique computes a class saliency map, specific to a given image and class. We show that such maps can be employed for weakly supervised object segmentation using classification ConvNets. Finally, we establish the connection between the gradient-based ConvNet visualisation methods and deconvolutional networks [Zeiler et al., 2013].
Subfields: Biophysics, Computer Vision and Pattern Recognition
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Biophysics
- Wojciech Samek
- K. Müller
- Grgoire Montavon
- Andrea Vedaldi
- Antonio Torralba
- Lior Wolf
- Bolei Zhou
- Been Kim
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Ming-Hsuan Yang
- Trevor Darrell
- Xiaogang Wang
- Jitendra Malik
- Andrew Zisserman
- Bernt Schiele
Citations in the corpus
- Neural Machine Translation in Linear Time (2016, methodology): “We further visualize the sensitivity of the ByteNet’s predictions to specific source and target inputs using gradient-based visualization (Simonyan et al., 2013).”
- Going deeper with convolutions (2014, methodology): “Recently, sparse representation based generative tracking methods have been developed for object tracking [1, 2, 3, 4, 5, 6, 7, 9, 10, 11].”
- Very Deep Convolutional Networks for Large-Scale Image Recognition (2014, neutral): “2 VGG-2013 [18] (weakly-supervised) 46.”
- What makes ImageNet good for transfer learning? (2016, methodology): “The pre-trained CNN is adapted by continued training on a target dataset, this process is referred to as finetuning.”
- Improving and Simplifying Pattern Exploiting Training (2021, background): “Future work could look into better masking strategies for labeled conditioned MLM, such as masking important tokens based on the the gradients of the logits for an example, as has been done for interpreting models (Simonyan et al., 2014).”
- Contrastive Learning of Medical Visual Representations from Paired Images and Text (2020, methodology): “To understand the representational difference that has led to this difference in performance, for all four initialization methods, we visualize in Figure 4 the saliency maps (Simonyan et al., 2014) corresponding to the correct class on sampled images from the CheXpert dataset.”
- ALICE: Active Learning with Contrastive Natural Language Explanations (2020, methodology): “We use the techniques in Simonyan et al. (2014) to visualze the saliency map.”
- Webly Supervised Learning of Convolutional Networks (2015, methodology): “The outline of our approach is shown in Figure 2.”
- Unsupervised Learning of Visual Representations by Solving Jigsaw Puzzles (2016, background): “Some recent work has devoted efforts towards the visualization of CNNs to better understand how they work and how we can exploit them [41,34,27,22].”
- Unrolled Generative Adversarial Networks (2016, background): “(14) This concept of backpropagating to generate images has been widely used in visualizing features from discriminative networks (Simonyan et al., 2013; Yosinski et al., 2015; Nguyen et al., 2016) and has been applied to explore the visual manifold of GANs in (Zhu et al.”
- Generating Images with Perceptual Similarity Metrics based on Deep Networks (2016, background): “Mahendran & Vedaldi (2015), as well as Simonyan et al. (2014) and Yosinski et al. (2015), apply gradient-based and Mahendran & Vedaldi (2015).”
- High Fidelity Visualization of What Your Self-Supervised Representation Knows About (2021, neutral): “Deterministic visualization methods: Many early works (Erhan et al., 2009; Zeiler & Fergus, 2014; Simonyan et al., 2013; Selvaraju et al., 2016; Smilkov et al., 2017) used gradient based techniques to visualize what is learned by neural networks.”
- SELFEXPLAIN: A Self-Explaining Architecture for Neural Text Classifiers (2021, methodology): “While post-hoc methods (Simonyan et al., 2014; Koh and Liang, 2017; Ribeiro et al., 2016) are often the only option 1 Code and data is publicly available at https:// github.com/dheerajrajagopal/SelfExplain”
- Measuring Association Between Labels and Free-Text Rationales (2020, methodology): “The gradient’s sign indicates the directionality of the change (Simonyan et al., 2014).”
- Semantic Image Inpainting with Deep Generative Models (2016, methodology): “Additionally, back-propagation has also been used to visualize and understand the learned features in a trained network, by “inverting” the network through updating the gradient at the input layer [26, 5, 35, 21].”
- Knowledge Neurons in Pretrained Transformers (2021, background): “The product of the gradients (of the output with respect to input features) and feature values is a reasonable base-line (Baehrens et al., 2010; Simonyan et al., 2014).”