VATUN: Visual Analytics for Testing and Understanding Convolutional Neural Networks

dc.contributor.authorPark, Cheonboken_US
dc.contributor.authorYang, Soyoungen_US
dc.contributor.authorNa, Inyoupen_US
dc.contributor.authorChung, Sunghyoen_US
dc.contributor.authorShin, Sungboken_US
dc.contributor.authorKwon, Bum Chulen_US
dc.contributor.authorPark, Deokgunen_US
dc.contributor.authorChoo, Jaegulen_US
dc.contributor.editorAgus, Marco and Garth, Christoph and Kerren, Andreasen_US
dc.date.accessioned2021-06-12T11:03:14Z
dc.date.available2021-06-12T11:03:14Z
dc.date.issued2021
dc.description.abstractConvolutional neural networks (CNNs) are popularly used in a wide range of applications, such as computer vision, natural language processing, and human-computer interaction. However, testing and understanding a trained model is difficult and very time-consuming. This is because their inner mechanisms are often considered as a 'black-box' due to difficulty in understanding the causal relationships between processes and results. To help the testing and understanding of such models, we present a user-interactive visual analytics system, VATUN, to analyze a CNN-based image classification model. Users can accomplish the following four tasks in our integrated system: (1) detect data instances in which the model confuses classification, (2) compare outcomes of the model by manipulating the conditions of the image, (3) understand reasons for the prediction of the model by highlighting highly influential parts from the image, and (4) analyze the overall what-if scenarios when augmenting the instances for each class. Moreover, by combining multiple techniques, our system lets users analyze behavior of the model from various perspectives. We conduct a user study of an image classification scenario with three domain experts. Our study will contribute to reducing the time cost for testing and understanding the CNN-based models in several industrial areas.en_US
dc.description.sectionheadersMachine Learning and SciVis Applications
dc.description.seriesinformationEuroVis 2021 - Short Papers
dc.identifier.doi10.2312/evs.20211047
dc.identifier.isbn978-3-03868-143-4
dc.identifier.pages7-11
dc.identifier.urihttps://doi.org/10.2312/evs.20211047
dc.identifier.urihttps://diglib.eg.org:443/handle/10.2312/evs20211047
dc.publisherThe Eurographics Associationen_US
dc.subjectHuman
dc.subjectcentered computing
dc.subjectVisualization toolkits
dc.titleVATUN: Visual Analytics for Testing and Understanding Convolutional Neural Networksen_US
Files
Original bundle
Now showing 1 - 2 of 2
Loading...
Thumbnail Image
Name:
007-011.pdf
Size:
5.29 MB
Format:
Adobe Portable Document Format
No Thumbnail Available
Name:
1043-file1.mp4
Size:
11.95 MB
Format:
Unknown data format
Collections