EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Hardware Especializado para Redes Neuronales: Una Revisión Sistemática

Daniel Gómez Cordioli

Revista Boaciencia. Negocios e Tecnología · 12.09.2023 · 3 (2) · 01-19

DOI: 10.59801/tyn.v3i2.224 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Revista Boaciencia. Negocios e Tecnología, online erschienen am 12.09.2023. Verfasst von Daniel Gómez Cordioli. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Neural networks have made great advances thanks to the evolution of hardware over the years, however, as the complexity of neural networks grows, so do their hardware requirements, as well as the time and cost for their deployment. This paper systematically reviews and discusses the use of specialized hardware for neural networks and their real-world applications. It is determined that specialized hardware in artificial intelligence and deep learning applications offers several significant advantages compared to the training and deployment of artificial intelligence applications. The most commonly used architectures were determined to be FPGAs, ASICs and ReRAMs. The major disadvantages were found to include limited bandwidth and data throughput. Parallelism, lower power consumption, reduced execution time, fault tolerance and cost optimization were identified as major advantages.

Zitieren

APA

Cordioli, D. G. (2023). Hardware Especializado para Redes Neuronales: Una Revisión Sistemática. Revista Boaciencia. Negocios e Tecnología, 3(2), 01-19. https://doi.org/10.59801/tyn.v3i2.224

Vancouver

Cordioli DG. Hardware Especializado para Redes Neuronales: Una Revisión Sistemática. Revista Boaciencia. Negocios e Tecnología. 2023;3(2):01-19. doi:10.59801/tyn.v3i2.224

BibTeX
@article{Cordioli2023,
  author = {Cordioli, Daniel Gómez},
  title = {Hardware Especializado para Redes Neuronales: Una Revisión Sistemática},
  journal = {Revista Boaciencia. Negocios e Tecnología},
  year = {2023},
  volume = {3},
  number = {2},
  pages = {01--19},
  doi = {10.59801/tyn.v3i2.224}
}
Kostenlos starten →