Answers are grounded in this paper and cite their sources.
Physical reservoir computing offers an energy-efficient hardware approach to temporal signal processing for edge artificial intelligence (AI). Memristors are promising candidates owing to their process compatibility and unique switching behaviors. Here, we developed an amorphous TiOx/AlOy superlattice-like (SLL) memristor that exhibits pulse-width-dependent volatility: as the pulse width decreases from 5 µs to 100 ns, the relaxation time constant of the conductance decreases from ~30 to ~6 µs, attributed to the relaxation of oxygen vacancies from the interfaces toward the TiOx layers. Using this fading memory, a reservoir model calibrated with the device predicts the second-order nonlinear autoregressive moving average (NARMA-2) task with a coefficient of determination (R²) of 0.94 and classifies heartbeat types from the MIT-BIH arrhythmia database with an accuracy of 84.1%. These results indicate that the SLL memristor is promising for low-power temporal signal processing in edge-AI applications.
No takes yet. Share an insight, caveat, or question.
Wang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: