To overcome the performance degradation in hardware neural networks (NNs) with non-ideal synapse devices, we proposed a novel neuromorphic architecture with both TiO x -based interfacial RRAM and CBRAM-based filamentary RRAM for highly accurate NN training and long-term inference reliability. We used a threshold-triggered training scheme, in which interfacial and filamentary RRAMs were programmed in a complementary fashion. This took advantage of the long retention time of the filamentary RRAM and the high-resolution, symmetric weight update in the interfacial RRAM. Additional evaluation of device parameters, such as linearity, precision, variation, and retention time, was conducted. An excellent pattern recognition accuracy of ~97% was achieved during training with the MNIST dataset. Thus, reliable inference accuracy after training was maintained using the filamentary RRAM.

中文翻译：

---

使用混合突触和补充训练的神经网络具有出色的模式识别精度


为了克服具有非理想突触设备的硬件神经网络（NN）的性能下降，我们提出了一种新颖的神经形态架构，具有基于 TiO x 的界面 RRAM 和基于 CBRAM 的丝状 RRAM，以实现高精度的 NN 训练和长期推理可靠性。我们使用了阈值触发的训练方案，其中界面和丝状 RRAM 以互补的方式进行编程。这利用了丝状 RRAM 的长保留时间和界面 RRAM 中的高分辨率、对称权重更新。对设备参数（例如线性度、精度、变化和保留时间）进行了额外评估。在使用 MNIST 数据集进行训练期间，获得了约 97% 的出色模式识别准确率。因此，使用丝状 RRAM 可以在训练后保持可靠的推理精度。