Ground truth based comparison of saliency maps algorithms
Abstract Deep neural networks (DNNs) have achieved outstanding results in domains such as image processing, computer vision, natural language processing and bioinformatics. In recent years, many methods have been proposed that can provide a visual explanation of decision made by such classifiers. Sa...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2023-10-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-023-42946-w |
| _version_ | 1797452868729438208 |
|---|---|
| author | Karolina Szczepankiewicz Adam Popowicz Kamil Charkiewicz Katarzyna Nałęcz-Charkiewicz Michał Szczepankiewicz Sławomir Lasota Paweł Zawistowski Krystian Radlak |
| author_facet | Karolina Szczepankiewicz Adam Popowicz Kamil Charkiewicz Katarzyna Nałęcz-Charkiewicz Michał Szczepankiewicz Sławomir Lasota Paweł Zawistowski Krystian Radlak |
| author_sort | Karolina Szczepankiewicz |
| collection | DOAJ |
| description | Abstract Deep neural networks (DNNs) have achieved outstanding results in domains such as image processing, computer vision, natural language processing and bioinformatics. In recent years, many methods have been proposed that can provide a visual explanation of decision made by such classifiers. Saliency maps are probably the most popular. However, it is still unclear how to properly interpret saliency maps for a given image and which techniques perform most accurately. This paper presents a methodology to practically evaluate the real effectiveness of saliency map generation methods. We used three state-of-the-art network architectures along with specially prepared benchmark datasets, and we proposed a novel metric to provide a quantitative comparison of the methods. The comparison identified the most reliable techniques and the solutions which usually failed in our tests. |
| first_indexed | 2024-03-09T15:14:52Z |
| format | Article |
| id | doaj.art-ac891ea96fa34318aec2041fcf852a11 |
| institution | Directory Open Access Journal |
| issn | 2045-2322 |
| language | English |
| last_indexed | 2024-03-09T15:14:52Z |
| publishDate | 2023-10-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj.art-ac891ea96fa34318aec2041fcf852a112023-11-26T13:11:18ZengNature PortfolioScientific Reports2045-23222023-10-0113111410.1038/s41598-023-42946-wGround truth based comparison of saliency maps algorithmsKarolina Szczepankiewicz0Adam Popowicz1Kamil Charkiewicz2Katarzyna Nałęcz-Charkiewicz3Michał Szczepankiewicz4Sławomir Lasota5Paweł Zawistowski6Krystian Radlak7Independent ResearcherDepartment of Electronics, Electrical Engineering and Microelectronics, Silesian University of TechnologyIndependent ResearcherInstitute of Computer Science, Warsaw University of TechnologyNVIDIADepartment of Electronics, Electrical Engineering and Microelectronics, Silesian University of TechnologyInstitute of Computer Science, Warsaw University of TechnologyInstitute of Computer Science, Warsaw University of TechnologyAbstract Deep neural networks (DNNs) have achieved outstanding results in domains such as image processing, computer vision, natural language processing and bioinformatics. In recent years, many methods have been proposed that can provide a visual explanation of decision made by such classifiers. Saliency maps are probably the most popular. However, it is still unclear how to properly interpret saliency maps for a given image and which techniques perform most accurately. This paper presents a methodology to practically evaluate the real effectiveness of saliency map generation methods. We used three state-of-the-art network architectures along with specially prepared benchmark datasets, and we proposed a novel metric to provide a quantitative comparison of the methods. The comparison identified the most reliable techniques and the solutions which usually failed in our tests.https://doi.org/10.1038/s41598-023-42946-w |
| spellingShingle | Karolina Szczepankiewicz Adam Popowicz Kamil Charkiewicz Katarzyna Nałęcz-Charkiewicz Michał Szczepankiewicz Sławomir Lasota Paweł Zawistowski Krystian Radlak Ground truth based comparison of saliency maps algorithms Scientific Reports |
| title | Ground truth based comparison of saliency maps algorithms |
| title_full | Ground truth based comparison of saliency maps algorithms |
| title_fullStr | Ground truth based comparison of saliency maps algorithms |
| title_full_unstemmed | Ground truth based comparison of saliency maps algorithms |
| title_short | Ground truth based comparison of saliency maps algorithms |
| title_sort | ground truth based comparison of saliency maps algorithms |
| url | https://doi.org/10.1038/s41598-023-42946-w |
| work_keys_str_mv | AT karolinaszczepankiewicz groundtruthbasedcomparisonofsaliencymapsalgorithms AT adampopowicz groundtruthbasedcomparisonofsaliencymapsalgorithms AT kamilcharkiewicz groundtruthbasedcomparisonofsaliencymapsalgorithms AT katarzynanałeczcharkiewicz groundtruthbasedcomparisonofsaliencymapsalgorithms AT michałszczepankiewicz groundtruthbasedcomparisonofsaliencymapsalgorithms AT sławomirlasota groundtruthbasedcomparisonofsaliencymapsalgorithms AT pawełzawistowski groundtruthbasedcomparisonofsaliencymapsalgorithms AT krystianradlak groundtruthbasedcomparisonofsaliencymapsalgorithms |