Results 251 to 260 of about 72,784 (298)

Feature Disentangling and Combination Implemented by Spin–Orbit Torque Magnetic Tunnel Junctions

open access: yesAdvanced Intelligent Systems, EarlyView.
Spin–orbit torque magnetic tunnel junctions (SOT‐MTJs) enable efficient feature disentangling and integration in image data. A proposed algorithm leverages SOT‐MTJs as true random number generators to disentangle and recombine features in real time, with experimental validation on emoji and facial datasets.
Xiaohan Li   +15 more
wiley   +1 more source

Nonplanar Atomic Layer Deposition (ALD)‐Niobium Oxide (NbOx) Neurons for Oscillatory Neural Network Applications

open access: yesAdvanced Intelligent Systems, EarlyView.
This article presents a nonplanar niobium oxide (NbOx) neuron device fabricated using an atomic layer deposition (ALD) method for use in oscillatory neural networks (ONNs). Potentially, such nonplanar geometry allows for high‐density arrays. The desired threshold switching (TS) characteristics are achieved through an interfacial method using a thin ...
Jaehyun Moon   +6 more
wiley   +1 more source

Compact Modeling of Volatile‐Switching Electrochemical Metallization Memory Cells by Means of the Electromotive Force

open access: yesAdvanced Intelligent Systems, EarlyView.
A volatile‐switching compact model of electrochemical metallization memory cells for neuromorphic architecture is developed and validated by reliable reproduction of device characterization measurements: I−V sweeps, SET kinetics, relaxation dynamics.
Rana Walied Ahmad   +4 more
wiley   +1 more source

QS4D: Quantization‐Aware Training for Efficient Hardware Deployment of Structured State‐Space Sequential Models

open access: yesAdvanced Intelligent Systems, EarlyView.
Quantization‐aware training creates resource‐efficient structured state space sequential S4(D) models for ultra‐long sequence processing in edge AI hardware. Including quantization during training leads to efficiency gains compared to pure post‐training quantization.
Sebastian Siegel   +5 more
wiley   +1 more source

A Fully Integrated Analog Processing‐in‐Memory System Based on Charge‐Trap Flash Synapse Arrays and Successive Integration‐and‐Rescaling Neurons

open access: yesAdvanced Intelligent Systems, EarlyView.
A fully integrated analog processing‐in‐memory system is demonstrated, combining charge‐trap flash synapse arrays with a successive integration‐and‐rescaling neuron circuit. The architecture performs bit‐sliced analog accumulation with high linearity and low power, achieving efficient and scalable analog in‐memory computing and bridging device‐level ...
Sojoong Kim   +4 more
wiley   +1 more source

Home - About - Disclaimer - Privacy