A Cluster of FPAAs to Recognize Images Using Neural Networks

Analog computing has been recovering its relevance in the recent years. Field-Programmable Analog Arrays (FPAAs) are the equivalent to Field-Programmable Gate Arrays (FPGAs) but in the analog and mixed-signal domain. In order to increase the amount of analog resources, in this brief a cluster of 40...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on circuits and systems. II, Express briefs Express briefs, 2021-11, Vol.68 (11), p.3391-3395
Main Authors: Garcia Moreno, Daniel, Del Barrio, Alberto A., Botella, Guillermo, Hasler, Jennifer
Format: Article
Language:English
Subjects:
analog computing
approximate computing
Artificial neural networks
classification
cluster
Clusters
Discrete cosine transforms
Field programmable analog arrays
Field programmable gate arrays
FPAA
Image classification
neural network
Neural networks
Software
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Analog computing has been recovering its relevance in the recent years. Field-Programmable Analog Arrays (FPAAs) are the equivalent to Field-Programmable Gate Arrays (FPGAs) but in the analog and mixed-signal domain. In order to increase the amount of analog resources, in this brief a cluster of 40 FPAAs is proposed. As a use case, a 19-8-6-4 feedforward Neural Network has been implemented on such cluster. With the help of a DCT-based software framework, this NN is able to classify 28 \times 28 images from MNIST. Results show that the analog network is able to obtain similar results as the software baseline network.
ISSN:1549-7747
1558-3791
DOI:10.1109/TCSII.2021.3077392