Adaptive Synaptic Programming in Differential Memristor Arrays
Dandao Xuebao/Journal of Ballistics · 30.06.2026 · 38 (3) · 108-118
DOI: 10.52783/dxjb.v38.496 ↗Kurz gefasst
Fachartikel in der Zeitschrift Dandao Xuebao/Journal of Ballistics, online erschienen am 30.06.2026. Verfasst von Mallela Deepthi und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Neuromorphic accelerators based on memristive crossbar arrays have emerged as promising platforms for implementing energy-efficient artificial neural networks due to their high integration density and in-memory computing capabilities. However, the inherent nonlinearity, device-to-device variability, conductance drift, and asymmetric switching characteristics of memristors pose significant challenges to achieving accurate and reliable synaptic weight programming. This paper proposes an Adaptive Conductance Programming (ACP) framework for differential memristor-based neuromorphic accelerators to improve programming precision and learning performance. The proposed approach dynamically adjusts programming pulse amplitude, width, and frequency based on real-time conductance feedback from paired differential memristor cells. By continuously monitoring conductance variations and compensating for device non-idealities, the ACP framework enables fine-grained weight updates and enhanced synaptic stability. The differential architecture further improves noise immunity and weight representation accuracy by encoding positive and negative synaptic weights across complementary memristor pairs. Simulation results demonstrate that the proposed method significantly reduces programming error, improves weight linearity, and enhances classification accuracy compared with conventional fixed-pulse programming techniques. Additionally, the framework achieves lower energy consumption and faster convergence during on-chip learning. These results indicate that adaptive conductance programming provides an effective solution for developing scalable, reliable, and energy-efficient neuromorphic accelerators for next-generation edge artificial intelligence applications.
Zitieren
APA
Deepthi, M., & Velpuri, V. (2026). Adaptive Synaptic Programming in Differential Memristor Arrays. Dandao Xuebao/Journal of Ballistics, 38(3), 108-118. https://doi.org/10.52783/dxjb.v38.496
Vancouver
Deepthi M, Velpuri V. Adaptive Synaptic Programming in Differential Memristor Arrays. Dandao Xuebao/Journal of Ballistics. 2026;38(3):108-118. doi:10.52783/dxjb.v38.496
BibTeX
@article{Deepthi2026,
author = {Deepthi, Mallela and Velpuri, Vijayalakshmi},
title = {Adaptive Synaptic Programming in Differential Memristor Arrays},
journal = {Dandao Xuebao/Journal of Ballistics},
year = {2026},
volume = {38},
number = {3},
pages = {108--118},
doi = {10.52783/dxjb.v38.496}
}
EUVIMED