A low-voltage, low-power reconfigurable current-mode softmax circuit for analog neural networks

Massimo Vatalaro, Marco Lanuzza, Felice Crupi, Tatiana Moposita, Lionel Trojman, Andrei Vladimirescu, Sebastiano Strangio

Producción científica: Contribución a una revistaArtículorevisión exhaustiva

7 Citas (Scopus)

Resumen

This paper presents a novel low-power low-voltage analog implementation of the softmax function, with electrically adjustable amplitude and slope parameters. We propose a modular design, which can be scaled by the number of inputs (and of corresponding outputs). It is composed of input current–voltage linear converter stages (1st stages), MOSFETs operating in a subthreshold regime implementing the exponential functions (2nd stages), and analog divider stages (3rd stages). Each stage is only composed of p-type MOSFET transistors. Designed in a 0.18 µm CMOS technology (TSMC), the proposed softmax circuit can be operated at a supply voltage of 500 mV. A ten-input/ten-output realization occupies a chip area of 2570 µm2 and consumes only 3 µW of power, representing a very compact and energy-efficient option compared to the corresponding digital implementations.

Idioma originalInglés
Número de artículo1004
PublicaciónElectronics (Switzerland)
Volumen10
N.º9
DOI
EstadoPublicada - 1 may. 2021
Publicado de forma externa

Huella

Profundice en los temas de investigación de 'A low-voltage, low-power reconfigurable current-mode softmax circuit for analog neural networks'. En conjunto forman una huella única.

Citar esto