“…metal–insulator–metal, MIM). , The full-hardware artificial neural networks (e.g., convolutional neural networks, CNNs) and low-power artificial dendrites with computing function have been realized by memristors. , In the past decades, various kinds of materials have been explored as the functional layer of memristors, such as the two-dimensional (2D) materials, − binary oxides, − binary nitrides, , organic materials, , halide perovskites, − etc . Among them, the halide perovskite has been considered as a promising candidate for memristor demonstrations owing to the favorable advantages in outstanding hysteresis effect, fast ion migration, and solution processability. ,, To date, perovskite memristors have achieved large resistance switching ratio (ON/OFF ratio, the ratio of high resistance to low resistance at the same voltage) of 10 9 , good cycle endurance of 10 6 , long retention time of 4.2 × 10 7 seconds, low power consumption of <5 × 10 –8 mW, and multilevel storage capabilities, which are comparable to or even better than rivals. ,, In addition, perovskite-based memristors have been applied in demonstrations of a logic gate circuit, − high density cross matrix memory, , and artificial synapse with neuromorphic computing. − …”