Memristor is a natural synapse because of its nanoscale and memory property, which influences the performance of memristive artificial neural networks. A three-variable memristor model is simplified with 15 kinds of properties, including the learning experience, the forgetting curve, the spiking time-dependent plasticity (STDP), the spiking rate dependent plasticity (SRDP), and the integration property. Through the analysis of the model, one more unobserved property called pseudo-polarity reversibility property is predicted by assuming the long-term memory is independent of memductance.

中文翻译：

---

回顾忆阻器特性

忆阻器是一种天然突触，因为它具有纳米级和记忆特性，影响忆阻人工神经网络的性能。将一个三变量忆阻器模型简化为具有学习经验、遗忘曲线、尖峰时间依赖可塑性（STDP）、尖峰速率依赖可塑性（SRDP）和积分特性等15种特性。通过对模型的分析，假设长时记忆与电感无关，预测了另一种未被观察到的特性，称为伪极性可逆性。