AdAM: Adaptive Fault-Tolerant Approximate Multiplier for Edge DNN Accelerators

Author:

Mahdi Taheri, Natalia Cherezova, Samira Nazari, Ahsan Rafiq, Ali Azarpeyvand, Tara Ghasempouri, Masoud Daneshtalab, Jaan Raik, Maksim Jenihhin

Keyword:

Computer Science, Artificial Intelligence, Artificial Intelligence (cs.AI), Hardware Architecture (cs.AR), Machine Learning (cs.LG)

journal:

date:

2024-03-05 00:00:00

Abstract

In this paper, we propose an architecture of a novel adaptive fault-tolerant approximate multiplier tailored for ASIC-based DNN accelerators.