A Learning-Rate Modulable and Reliable TiOx Memristor Array for Robust, Fast, and Accurate Neuromorphic Computing

Realization of memristor-based neuromorphic hardware system is important to achieve energy efficient bigdata processing and artificial intelligence in integrated device system-level. In this sense, uniform and reliable titanium oxide (TiOx) memristor array devices are fabricated to be utilized as constituent device element in hardware neural network, representing passive matrix array structure enabling vector-matrix multiplication process between multisignal and trained synaptic weight. In particular, in situ convolutional neural network hardware system is designed and implemented using a multiple 25 x 25 TiOx memristor arrays and the memristor device parameters are developed to bring global constant voltage programming scheme for entire cells in crossbar array without any voltage tuning peripheral circuit such as transistor. Moreover, the learning rate modulation during in situ hardware training process is successfully achieved due to superior TiOx memristor performance such as threshold uniformity (approximate to 2.7%), device yield (> 99%), repetitive stability (approximate to 3000 spikes), low asymmetry value of approximate to 1.43, ambient stability (6 months), and nonlinear pulse response. The learning rate modulable fast-converging in situ training based on direct memristor operation shows five times less training iterations and reduces training energy compared to the conventional hardware in situ training at approximate to 95.2% of classification accuracy.

Automated summary

Realization of memristor-based neuromorphic hardware system is important for energy efficient big data processing and artificial intelligence. Uniform and reliable titanium oxide (TiOx) memristor array devices are fabricated to enable vector-matrix multiplication process in a hardware neural network. A convolutional neural network hardware system using TiOx memristor arrays is designed and implemented, achieving learning rate modulation and fast convergence. This in situ training reduces training iterations and energy consumption while maintaining high classification accuracy.