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...

Full beskrivning

Bibliografiska uppgifter

Huvudupphov:	Georgios Zervakis, Fabio Frustaci, Ourania Spantidi, Iraklis Anagnostopoulos, Hussam Amrouch, Jorg Henkel
Materialtyp:	Artigo
Språk:	Inglês
Utgiven:	IEEE 2025-01-01
Serie:	IEEE Access
Ämnen:	Approximate computing approximate multipliers control variate deep neural networks error correction low power
Länkar:	https://ieeexplore.ieee.org/document/10924159/
Taggar:	Lägg till en tagg Inga taggar, Lägg till första taggen!