A Taxonomy and Library for Visualizing Learned Features in Convolutional Neural Networks
By Felix Grün, Christian Rupprecht, Nassir Navab, Federico Tombari. 2016
77 citations recorded by the source corpus.
Over the last decade, Convolutional Neural Networks (CNN) saw a tremendous surge in performance. However, understanding what a network has learned still proves to be a challenging task. To remedy this unsatisfactory situation, a number of groups have recently proposed different methods to visualize the learned models. In this work we suggest a general taxonomy to classify and compare these methods, subdividing the literature into three main categories and providing researchers with a terminology to base their works on. Furthermore, we introduce the FeatureVis library for MatConvNet: an extendable, easy to use open source library for visualizing CNNs. It contains implementations from each of the three main classes of visualization methods and serves as a useful tool for an enhanced understanding of the features learned by intermediate layers, as well as for the analysis of why a network might fail for certain examples.
Subfields: Artificial Intelligence, 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 Artificial Intelligence
- Willem F. G. Haselager
- Marcel A. J. van Gerven
- Mao Ye
- Christin Seifert
- Liu Ren
- Aisha Aamir
- Abdelouahab Moussaoui
- Dhruvi Jain
In Computer Vision and Pattern Recognition
- Xiaoming Liu
- Amin Jourabloo
- Christian Rupprecht
- Luc Van Gool
- C. Pal
- Xiaogang Wang
- Andrew Zisserman
- Marcel A. J. van Gerven
Citations in the corpus
- Explanation Methods in Deep Learning: Users, Values, Concerns and Challenges (2018, methodology): “However, apart from a non-exhaustive overview of existing methods (Montavon et al., 2018) and classification schemes for purely visual methods (Gr¨un et al., 2016; Seifert et al., 2017; Zeng, 2016; Kindermans et al., 2017), little is known about efforts to rigorously map the landscape of explanation…”
- Deep Learning for Tomato Diseases: Classification and Symptoms Visualization (2017, background): “In deep learning, many studies proposed the visualization of learned features, in order to explain how classifier gives the final results and how features are constructed (Grün et al. 2016; Yosinski et al. 2015; Zeiler and Fergus 2014).”
- Finding and Visualizing Weaknesses of Deep Reinforcement Learning Agents (2019, background): “As outlined in [12], we differentiate between saliency methods , that highlight decision-relevant regions given an input image, methods that synthesize an image (pre-image) that fulfills a certain criterion, such as activation maximization [9] or input reconstruction , and methods that are…”
- Visual Analytics in Deep Learning: An Interrogative Survey for the Next Frontiers (2018, unknown): “For more details about these techniques, such as input modification, deconvolutional methods [10], and input reconstruction methods, see the following taxonomies and surveys for visualizing learned features in CNNs [22], [89], [90].”
- Do Convolutional Neural Networks Learn Class Hierarchy? (2017, methodology): “These maps are computed using the FeatureVis library [24] and the MatConvNet toolbox [69].”
- Visualizations of Deep Neural Networks in Computer Vision: A Survey (2017, methodology): “Additionally, we categorize the papers according to the taxonomy of Grün [25] (input modification methods, de-convolutional methods and input reconstruction) if this taxonomy is applicable.”