Experiments on concept localization on XTRAINS and SCDB datasets. Custom network, ResNet-18 and MobileNet-V2.
The repository contains the results of experiments on the localization of ontology concepts in deep convolutional neural networks of three different architectures (Custom, MobileNet-V2, and ResNet-18).
Experiments were conducted on two datasets:
- Explainable Abstract Trains Dataset (XTRAINS)
Associated paper:
M. de S. Ribeiro, L. Krippahl, and J. Leite, “Explainable Abstract Trains Dataset,” Dec. 2020, [Online]. Available: http://arxiv.org/abs/2012.12115 - Simple Concept DataBase (SCDB)
Associated paper:
A. Lucieri, M. N. Bajwa, A. Dengel, and S. Ahmed, “Explaining AI-Based Decision Support Systems Using Concept Localization Maps,” Commun. Comput. Inf. Sci., vol. 1332, pp. 185–193, May 2020, doi: 10.1007/978-3-030-63820-7_21.