Multimodal Side- Tuning for Document Classification - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

Multimodal Side- Tuning for Document Classification

Résumé

In this paper, we propose to exploit the side-tuning framework for multimodal document classification. Side-tuning is a methodology for network adaptation recently introduced to solve some of the problems related to previous approaches. Thanks to this technique it is actually possible to overcome model rigidity and catastrophic forgetting of transfer learning by fine- tuning. The proposed solution uses off-the-shelf deep learning architectures leveraging the side-tuning framework to combine a base model with a tandem of two side networks. We show that side-tuning can be successfully employed also when different data sources are considered, e.g. text and images in document classification. The experimental results show that this approach pushes further the limit for document classification accuracy with respect to the state of the art.

Dates et versions

hal-04370806 , version 1 (03-01-2024)

Licence

Domaine public

Identifiants

Citer

Stefano Pio Zingaro, Giuseppe Lisanti, Maurizio Gabbrielli. Multimodal Side- Tuning for Document Classification. 2020 25th International Conference on Pattern Recognition (ICPR), Jan 2021, Milan, Italy. pp.5206-5213, ⟨10.1109/ICPR48806.2021.9413208⟩. ⟨hal-04370806⟩
6 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More