Title :
Programmable non-linearity for neural networks applications
Author :
Sargeni, Fausto ; Bonaiuto, Vincenzo
Author_Institution :
Dept. of Electron. Eng., Univ. of Rome Tor Vergata, Rome, Italy
Abstract :
The implementation of analogue circuits for modeling Artificial Neural Networks as well as Neuromorphic architectures, makes wide use of nonlinear circuits where the programmability feature could be a very interesting characteristic. This paper deals with the design of a "current-mode" digitally programmable transconductance comparator. In particular, it has been properly tailored for a "time-division architecture" implementation of a first order STAR CNN system.
Keywords :
analogue circuits; comparators (circuits); current-mode circuits; neural nets; nonlinear network synthesis; analogue circuits; artificial neural networks; current-mode transconductance amplifier; digitally programmable transconductance comparator; first order STAR CNN system; neuromorphic architectures; nonlinear circuits; programmable nonlinearity; time-division architecture; Neural networks;
Conference_Titel :
Circuits and Systems, 2009. MWSCAS '09. 52nd IEEE International Midwest Symposium on
Conference_Location :
Cancun
Print_ISBN :
978-1-4244-4479-3
Electronic_ISBN :
1548-3746
DOI :
10.1109/MWSCAS.2009.5235907
