A flexible artificial synapse based on the two-dimensional CuInS2 memristor for neural morphology calculation

Neuromorphic computing , consisting of artificial synapses and neural network algorithms, is much more energy efficient and is expected to replace traditional von Neumann computer architectures in the development of next-generation intelligent systems. Therefore, in this work, the flexible Au/CuInS 2 /Cu memristor were prepared using a low-temperature solution method. The flexible memristor demonstrates notable bipolar resistance switching performance, including a smaller threshold voltage (∼0.6 V), a larger switching ratio (∼10 2 ) and lower power consumption (∼10 nW). In addition, the device accurately simulates simple biological synaptic behaviors, such as long-term and short-term plasticity (LTP and STP), paired-pulse facilitation (PPF) and spike-time-dependent plasticity (STDP). The MNIST-based Convolutional Neural Network (CNN) architecture was created to apply an enhanced stochastic adaptive algorithm and raise the recognition accuracy of the MNIST and Fashion-MNIST datasets to 96.7 % and 81.4 %, respectively. The results suggest that the 2D CuInS 2 nanomaterials with outstanding performance provide a viable method for developing future flexible neural morphology computing systems.