Open access
Sep 2026
Activation function saturation determines robustness to conductance nonlinearity in analog in-memory learning
This paper compares five activation functions on MNIST and Fashion-MNIST using an idealized software configuration and a hardware configuration parameterized by the measured conductance characteristics of a fabricated TiO 2 device, and inverts the one familiar from digital deep learning.
Baki Gökgöz, Tolga Aydın, F. Gül
· Scientific Reports · 0 citations