fig14
Figure 14. (A) Asymmetric anti-Hebbian configuration, (B) Symmetric Hebbian configuration, (C) Symmetric anti-Hebbian configuration realized by the device. (D) Schematic diagram of the multilayer neural network used for recognition. (E and F) Weight distribution and accuracy of software training and conductance maps[70]. Reprinted from Ref.[70], Copyright (2022), with permission from Elsevier.
