Cargando…
Compact Neural Architecture Designs by Tensor Representations
We propose a framework of tensorial neural networks (TNNs) extending existing linear layers on low-order tensors to multilinear operations on higher-order tensors. TNNs have three advantages over existing networks: First, TNNs naturally apply to higher-order data without flattening, which preserves...
Autores principales: | Su, Jiahao, Li, Jingling, Liu, Xiaoyu, Ranadive, Teresa, Coley, Christopher, Tuan, Tai-Ching, Huang, Furong |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Frontiers Media S.A.
2022
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8959219/ https://www.ncbi.nlm.nih.gov/pubmed/35355829 http://dx.doi.org/10.3389/frai.2022.728761 |
Ejemplares similares
-
Decoding of the neural representation of the visual RGB color model
por: Wu, Yijia, et al.
Publicado: (2023) -
A tale of two lexica: Investigating computational pressures on word representation with neural networks
por: Avcu, Enes, et al.
Publicado: (2023) -
An Explainable Multimodal Neural Network Architecture for Predicting Epilepsy Comorbidities Based on Administrative Claims Data
por: Linden, Thomas, et al.
Publicado: (2021) -
Strategy to improve the accuracy of convolutional neural network architectures applied to digital image steganalysis in the spatial domain
por: Tabares-Soto, Reinel, et al.
Publicado: (2021) -
Image Completion in Embedded Space Using Multistage Tensor Ring Decomposition
por: Sedighin, Farnaz, et al.
Publicado: (2021)