Fusion of evidential CNN classifiers for image classification - Archive ouverte HAL Access content directly
Conference Papers Year : 2021

Fusion of evidential CNN classifiers for image classification

(1) , (1) , (1, 2)
1
2

Abstract

We propose an information-fusion approach based on belief functions to combine convolutional neural networks. In this approach, several pre-trained DS-based CNN architectures extract features from input images and convert them into mass functions on different frames of discernment. A fusion module then aggregates these mass functions using Dempster's rule. An end-to-end learning procedure allows us to fine-tune the overall architecture using a learning set with soft labels, which further improves the classification performance. The effectiveness of this approach is demonstrated experimentally using three benchmark databases.
Fichier principal
Vignette du fichier
R1-clean-E-fusion-DL_TD.pdf (394.7 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03511144 , version 1 (04-01-2022)

Identifiers

Cite

Zheng Tong, Philippe Xu, Thierry Denœux. Fusion of evidential CNN classifiers for image classification. 6th International Conference on Belief Functions (BELIEF 2021), Oct 2021, Shanghai, China. pp.168-176, ⟨10.1007/978-3-030-88601-1_17⟩. ⟨hal-03511144⟩
19 View
20 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More