Energy-Efficient Machine Learning Algorithms - Ecole Nationale d'Ingénieurs de Brest Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

Energy-Efficient Machine Learning Algorithms

Résumé

When designing electronic systems, a standard technique to reduce the energy consumption consists of agressively downscaling the voltage supply. However, due to physical limitations, further reducing the power supply of next generations of electronic devices will make computational units unreliable, which may introduce faults in the computation operations realized on these chips [3]. On the other hand, tolerating faults in the computation operations gives us the opportunity to address a tradeoff between algorithm performance and energy consumption. This is the issue we consider in this talk.
Tinkos19_Dupraz.pdf (81.46 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-02925052 , version 1 (02-09-2020)

Identifiants

  • HAL Id : hal-02925052 , version 1

Citer

Elsa Dupraz, Lav Varshney. Energy-Efficient Machine Learning Algorithms. The 7th Conference on Information Theory and Complex Systems (TINKOS 2019), Oct 2019, Belgrade, Serbia. ⟨hal-02925052⟩
58 Consultations
17 Téléchargements

Partager

Gmail Facebook X LinkedIn More