Multi-Exit Resource-Efficient Neural Architecture for Image Classification with Optimized Fusion Block - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

Multi-Exit Resource-Efficient Neural Architecture for Image Classification with Optimized Fusion Block

Résumé

In this paper, we propose a test-time resource-efficient neural architecture for image classification. Building on MSDNet [12], our multi-exit architecture excels in both anytime classification, which allows progressive updates of predictions for test examples and facilitates early output, and budgeted batch classification, which allows flexible allocation of computational resources across inputs to classify a set of examples within a fixed budget. Our proposed multi-exit architecture achieves state-of-the-art performance on CIFAR10 and CIFAR100 in these two critical scenarios, thanks to a novel feature fusion building block combined with an efficient stem block.
Fichier principal
Vignette du fichier
iccvw-multu-exit.pdf (214.33 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04181149 , version 1 (15-08-2023)
hal-04181149 , version 2 (18-08-2023)

Identifiants

  • HAL Id : hal-04181149 , version 2

Citer

Youva Addad, Alexis Lechervy, Frédéric Jurie. Multi-Exit Resource-Efficient Neural Architecture for Image Classification with Optimized Fusion Block. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, Oct 2023, Paris, France. pp.1486-1491. ⟨hal-04181149v2⟩
107 Consultations
48 Téléchargements

Partager

Gmail Facebook X LinkedIn More