25 April 2025
Memristive nanowire-based neurons offer a novel and promising approach to neuromorphic computing. Leveraging the unique properties of memristors-electronic components whose resistance changes based on the history of applied voltage, these neurons emulate the dynamics of biological systems, particularly synaptic plasticity.
Memristors’ inherent ability to “remember” past electrical states makes them ideal candidates for simulating temporally dynamic processes such as short-term plasticity, a cornerstone of adaptive neural computation. In neuromorphic systems, memristors are gaining traction as efficient analog components for replicating the complex dynamics of biological memory. However, to integrate the realism of their continuous-time physics into digital Artificial Neural Networks (ANNs), discrete-time modeling is essential. Such adaptation bridges the gap between analog hardware and digital frameworks, enabling memristor-inspired neurons to be deployed within existing ANN architectures.
In this work, EMERGE partners from the University of Pisa introduce a novel discrete-time model for a neural processing unit based on the physical principles of memristive nanowires. The model approximates the conductance dynamics of memristors and casts them as artificial neuron dynamics, making it suitable for digital systems.
Read the paper in the link below.
More Information
