Very fast programmable CNN based on FG-inverter

In this paper, a different CNN implementation based on floating-gate inverters is presented. The methodology adjusts the original CNN model in order to incorporate a floating-gate inverter and reduce hardware complexity. The approach brings a fast response compatible with standard digital circuits....

Full description

Saved in:
Bibliographic Details
Main Authors: Molinar-Solis, J.K., Gomez-Castenada, F., Moreno-Cadenas, J.A., Ponce-Ponce, V.H.
Format: Conference Proceeding
Language:English
Subjects:
Capacitance
Cellular neural networks
Clocks
Energy consumption
Hardware
Image processing
Inverters
MOSFETs
Switches
Voltage
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, a different CNN implementation based on floating-gate inverters is presented. The methodology adjusts the original CNN model in order to incorporate a floating-gate inverter and reduce hardware complexity. The approach brings a fast response compatible with standard digital circuits. Eight different processing tasks were tested on a 4times4 CNN prototype fabricated on AMI 1.2 micron through MOSIS
ISSN:0271-4302
2158-1525
DOI:10.1109/ISCAS.2006.1693437