QR code

Leveraging Highly Approximated Multipliers in DNN Inference

In this work, we present our control variate approximation technique that enables the exploitation of highly approximate multipliers in Deep Neural Network (DNN) accelerators. Our approach does not require retraining and significantly decreases the induced error due to approximate multiplications, i...

Volledige beschrijving

Bewaard in:
Hoofdauteurs: Georgios Zervakis, Fabio Frustaci, Ourania Spantidi, Iraklis Anagnostopoulos, Hussam Amrouch, Jorg Henkel
Formaat: Artigo
Taal:Inglês
Gepubliceerd in: IEEE 2025-01-01
Reeks:IEEE Access
Onderwerpen:
Approximate computing
approximate multipliers
control variate
deep neural networks
error correction
low power
Online toegang:https://ieeexplore.ieee.org/document/10924159/
Tags: Voeg label toe
Geen labels, Wees de eerste die dit record labelt!

Gelijkaardige items