* If you want to update the article please login/register
A classic von Neumann computation system based on memristor arrays holds the promise to address the speed and energy limitations of the classical von Neumann computation system. However, the stochasticity of ions transport in conventional oxide-based memristor is causing significant intrinsic variability, which compromises learning efficiency and hinders the development of neural network hardware accelerators. In addition, we continue to perform convolutional image processing using various crossbar kernels that achieved a high recognition accuracy of 93. 4% thanks to the highly linear and symmetric analogue weight update, as well as several conductance states, demonstrating its capability beyond von Neumann's computation.
* Please keep in mind that all text is summarized by machine, we do not bear any responsibility, and you should always check original source before taking any actions