作者：Chen Peng, Lin Peng, Pan Gang；等
来源：NATURE COMMUNICATIONS 卷:14 期:1 页(号):6184 出版时间: OCT 2023
Learning-on-chip is a highly desired computing technology that directly update the hardware-based synaptic weights for training neural networks. We utilized an electrochemical device as artificial synapse that can be programmed from one state to the other without any feedback adjustments. We showed, for the first time, that such open-loop programmability can be experimentally achieved in arrays, which offers a unique solution to accelerate training tasks.
