Results 91 to 100 of about 127,617 (265)

Unbiasing Truncated Backpropagation Through Time

, 2017
Truncated Backpropagation Through Time (truncated BPTT) is a widespread method for learning recurrent computational graphs. Truncated BPTT keeps the computational benefits of Backpropagation Through Time (BPTT) while relieving the need for a complete ...
Ollivier, Yann, Tallec, Corentin
core  

Water‐Assisted Exfoliation of HfO2‐Based Membrane for Flexible Robust Ferroelectric Synaptic Transistors

Advanced Science, EarlyView.
A flexible freestanding HfO2‐based ferroelectric membrane is achieved via a water‐assisted exfoliation technique using a Sr4Al2O₇ sacrificial layer. The BaTiO3/Hf0.5Zr0.5O2/BaTiO3 heterostructure maintains robust ferroelectricity and exhibits reliable synaptic plasticity.
Han Zhang, Miao Zeng, Jingxin Chen, Lei Wang, Sai Jiang, Qian Cao, Xuebo Li, Zekun Hou, Chun Shen, Yulin Gan, Ning Fang, Zhaoliang Liao, Ziyao Zhou, Lin Hao   +13 more
wiley   +1 more source

Multi‐Physical Field Modulated P‐Bit Device Based on VO2 Thin Film

Advanced Science, EarlyView.
We have proposed a VO2‐based P‐bit device where synergistic multi‐physical field modulation enables real‐time tunability of randomness. Besides introducing a new phase‐change material‐based device approach for high‐performance P‐bits, this study also demonstrates a synergistic multi‐physical field modulation strategy that opens new opportunities for ...
Bowen Sun, Jianjun Li, Ting Zhou, Jinglin Zhu, Meiling Liu, Zhihan Lin, Chang Wang, Chengyu Li, Yingxue Chen, Xiaokun Huo, Chongwen Zou   +10 more
wiley   +1 more source

Unsupervised end-to-end training with a self-defined target

Neuromorphic Computing and Engineering
Designing algorithms for versatile AI hardware that can learn on the edge using both labeled and unlabeled data is challenging. Deep end-to-end training methods incorporating phases of self-supervised and supervised learning are accurate and adaptable to
Dongshu Liu, Jérémie Laydevant, Adrien Pontlevy, Damien Querlioz, Julie Grollier   +4 more
doaj   +1 more source

Physical Implementation of Optical Material‐Based Neural Networks Processing Enabled by Long‐Persistent Luminescence

Advanced Science, EarlyView.
This study reports on the physical implementation of optical material‐based neural processing using long‐persistent luminescence as memory‐retention and nonlinear optical material. The system performs optical‐domain preprocessing with opto‐electronic interfaces for stimulus delivery and readout, enabling real‐time demonstrations including Pong gameplay
Sangwon Wi, Yunsang Lee
wiley   +1 more source

All Organic Fully Integrated Neuromorphic Crossbar Array

Advanced Electronic Materials, EarlyView.
In this work, the first fully integrated crossbar array of electrochemical random‐access memory (ECRAM) that is composed entirely of organic materials is represented. This array can perform inference and in situ parallel training and is capable of classifying linearly separable 2D and 3D classification tasks with high accuracy.
Setareh Kazemzadeh, Tim Stevens, Yoeri van de Burgt   +2 more
wiley   +1 more source

Optical Charge Trap Memory Based on Graphene/ZnO Heterostructures for Long‐Term Retention and Adaptive Learning

Advanced Electronic Materials, EarlyView.
A biocompatible graphene/ZnO optical charge trap memory (CTM) is reported with over 54 h retention, enabled by interfacial photodoping. Using transient absorption spectroscopy and electrical analysis, charge transfer quenching is elucidated and reveal that a large energy barrier at the interface is responsible for long‐term memory retention.
Seungmin Shin, Junhyung Kim, Hyungdoh Lee, Yixuan Dou, Jacob T. Heiden, Seung Eun Lee, Jinwoo Song, Lina Quan, Giulia Tagliabue, Min Seok Jang, Himchan Cho   +10 more
wiley   +1 more source

Organic Thin‐Film Transistors for Neuromorphic Computing

Advanced Electronic Materials, EarlyView.
Organic thin‐film transistors (OTFTs) are reviewed for neuromorphic computing applications, highlighting their power‐efficient, and biological time‐scale operation. This article surveys OFET and OECT devices, compares them with memristors and CMOS, analyzes how fabrication parameters shape spike‐based metrics, proposes standardized characterization ...
Luke McCarthy, Mohan V. Jacob, Mostafa Rahimi Azghadi   +2 more
wiley   +1 more source

Tropical Backpropagation

, 2023
This work introduces tropicalization, a novel technique that delivers tropical neural networks as tropical limits of deep ReLU networks. Tropicalization transfers the initial weights from real numbers to those in the tropical semiring while maintain- ing the underlying graph of the network.
Ceyhan, Özgür, Lucchetti, Federico
openaire   +1 more source

Extended backpropagation algorithm

Electronics Letters, 1993
A fast learning algorithm for feedforward neural networks is described. This algorithm is based on modifications to the backpropagation algorithm. The learning rate and momentum coefficient are adapted according to the coefficient of correlation between the downhill gradient and the previous weight update.
Y.F. Yam, T.W.S. Chow
openaire   +1 more source