An Information-theoretic Visual Analysis Framework for Convolutional Neural Networks

Simple item page
dc.contributor.authorShen, Jingyien_US
dc.contributor.authorShen, Han-Weien_US
dc.contributor.editorFrosini, Patrizio and Giorgi, Daniela and Melzi, Simone and Rodolà, Emanueleen_US
dc.date.accessioned2021-10-25T11:53:44Z
dc.date.available2021-10-25T11:53:44Z
dc.date.issued2021
dc.description.abstractDespite the great success of Convolutional Neural Networks (CNNs) in Computer Vision and Natural Language Processing, the working mechanism behind CNNs is still under extensive discussion and research. Driven by strong demand for the theoretical explanation of neural networks, some researchers utilize information theory to provide insight into the black-box model. However, to the best of our knowledge, employing information theory to quantitatively analyze and qualitatively visualize neural networks has not been extensively studied in the visualization community. In this paper, we combine information entropies and visualization techniques to shed light on how CNN works. Specifically, we first introduce a data model to organize the data that can be extracted from CNN models. Then we propose two ways to calculate entropy under different circumstances. To provide a fundamental understanding of the basic building blocks of CNNs (e.g., convolutional layers, pooling layers, normalization layers) from an information-theoretic perspective, we develop a visual analysis system, CNNSlicer. CNNSlicer allows users to interactively explore the amount of information changes inside the model. With case studies on the widely used benchmark datasets (MNIST and CIFAR-10), we demonstrate the effectiveness of our system in opening the black-box of CNNs.en_US
dc.description.sectionheadersVisualization
dc.description.seriesinformationSmart Tools and Apps for Graphics - Eurographics Italian Chapter Conference
dc.identifier.doi10.2312/stag.20211486
dc.identifier.isbn978-3-03868-165-6
dc.identifier.issn2617-4855
dc.identifier.pages163-173
dc.identifier.urihttps://doi.org/10.2312/stag.20211486
dc.identifier.urihttps://diglib.eg.org:443/handle/10.2312/stag20211486
dc.publisherThe Eurographics Associationen_US
dc.subjectHuman
dc.subjectcentered computing → Visual analytics
dc.titleAn Information-theoretic Visual Analysis Framework for Convolutional Neural Networksen_US
Files
Original bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
163-173.pdf
Size:
3.6 MB
Format:
Adobe Portable Document Format
Download
Collections
Italian Chapter Conference 2021 - Smart Tools and Apps in Graphics