Mitigating Non-idealities of Memristive-based Artificial Neural Networks - an Algorithmic Approach

The computing power demands to run artificial neural networks (ANNs) are increasing at rates much greater than improvements made with current CMOS-based technologies. The demand has contributed to a need for novel paradigms, including memristor-based accelerators. This work explores two algorithmic...

Full description

Saved in:
Bibliographic Details
Published in2022 6th IEEE Electron Devices Technology & Manufacturing Conference (EDTM) pp. 399 - 401
Main Authors Mehonic, Adnan, Joksas, Dovydas, Barmpatsalos, Nikolaos, Ng, Wing H., Kenyon, Anthony J., Wang, Erwei, Constantinides, George
Format Conference Proceeding
LanguageEnglish
Published IEEE 06.03.2022
Subjects
Artificial neural networks
deep-learning accelerator
energy-efficiency
Inference algorithms
Manufacturing
memristor
Memristors
neuromorphic
Power demand
Training
Online AccessGet full text

Cover

More Information
Summary:The computing power demands to run artificial neural networks (ANNs) are increasing at rates much greater than improvements made with current CMOS-based technologies. The demand has contributed to a need for novel paradigms, including memristor-based accelerators. This work explores two algorithmic approaches to mitigate non-idealities inherent in most memristor-based systems. The first is to apply a concept of committee machines during inference, and the second is nonideality-aware training of memristor-based ANNs.
DOI:10.1109/EDTM53872.2022.9798334