Centre de documentation
scientifique

Structured binary neural networks for image recognition / Bohan Zhuang in International journal of computer vision, vol 130 n° 9 (September 2022)  

Public [article]  inInternational journal of computer vision > vol 130 n° 9 (September 2022) . - pp 2081 - 2102 Titre : Structured binary neural networks for image recognition Type de document : Article/Communication Auteurs : Bohan Zhuang, Auteur ; Chunhua Shen, Auteur ; Mingkui Tan, Auteur ; et al., Auteur Année de publication : 2022 Article en page(s) : pp 2081 - 2102 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] apprentissage profond [Termes IGN] décomposition [Termes IGN] détection d'objet [Termes IGN] implémentation (informatique) [Termes IGN] logique binaire [Termes IGN] segmentation sémantique Résumé : (auteur) In this paper, we propose to train binarized convolutional neural networks (CNNs) that are of significant importance for deploying deep learning to mobile devices with limited power capacity and computing resources. Previous works on quantizing CNNs often seek to approximate the floating-point information of weights and/or activations using a set of discrete values. Such methods, termed value approximation here, typically are built on the same network architecture of the full-precision counterpart. Instead, we take a new “structured approximation” view for network quantization — it is possible and valuable to exploit flexible architecture transformation when learning low-bit networks, which can achieve even better performance than the original networks in some cases. In particular, we propose a “group decomposition” strategy, termed GroupNet, which divides a network into desired groups. Interestingly, with our GroupNet strategy, each full-precision group can be effectively reconstructed by aggregating a set of homogeneous binary branches. We also propose to learn effective connections among groups to improve the representation capability. To improve the model capacity, we propose to dynamically execute sparse binary branches conditioned on input features while preserving the computational cost. More importantly, the proposed GroupNet shows strong flexibility for a few vision tasks. For instance, we extend the GroupNet for accurate semantic segmentation by embedding the rich context into the binary structure. The proposed GroupNet also shows strong performance on object detection. Experiments on image classification, semantic segmentation, and object detection tasks demonstrate the superior performance of the proposed methods over various quantized networks in the literature. Moreover, the speedup and runtime memory cost evaluation comparing with related quantization strategies is analyzed on GPU platforms, which serves as a strong benchmark for further research. Numéro de notice : A2022-637 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1007/s11263-022-01638-0 Date de publication en ligne : 22/06/2022 En ligne : https://doi.org/10.1007/s11263-022-01638-0 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=101443 [article]