Results 61 to 70 of about 25,600 (268)

Nonlinear backpropagation: doing backpropagation without derivatives of the activation function [PDF]

IEEE Transactions on Neural Networks, 1997
The conventional linear backpropagation algorithm is replaced by a nonlinear version, which avoids the necessity for calculating the derivative of the activation function. This may be exploited in hardware realizations of neural processors. In this paper we derive the nonlinear backpropagation algorithms in the framework of recurrent backpropagation ...
John A. Hertz, Anders Krogh, Benny Lautrup, Torsten Lehmann   +3 more
openaire   +2 more sources

Triple‐Mode Ferroelectric Thin‐Film Transistor for Hybrid Electrical–Optical Reservoir Computing

Advanced Science, EarlyView.
A triple‐mode ferroelectric thin‐film transistor is developed by integrating Si3N4/HZO/IGZO layers to realize three independent memory modes: electric long‐term, electric short‐term, and optical short‐term. This single‐device architecture functions as both a reservoir and readout layer, achieving 92.43% MNIST accuracy. It offers a fully hardware‐based,
Hyeonho Lee, Seungjun Lee, Tae‐Hyeon Kim, Yonghee Jeong, Gwangmin An, Minsu Ko, Junsu Yu, Heung Soo Kim, Yang Chai, Sungjun Kim   +9 more
wiley   +1 more source

A Novel Neuro-Fuzzy Model for Multivariate Time-Series Prediction

Data, 2018
Time series forecasting can be a complicated problem when the underlying process shows high degree of complex nonlinear behavior. In some domains, such as financial data, processing related time-series jointly can have significant benefits.
Alexander Vlasenko, Nataliia Vlasenko, Olena Vynokurova, Dmytro Peleshko   +3 more
doaj   +1 more source

Interpretable Machine Learning Framework for Nb─Si Based Alloy Design with Enhanced Fracture Toughness

Advanced Science, EarlyView.
An interpretable machine learning framework integrating SHAP and PDP analysis identifies critical design descriptors from 139 physicochemical features for Nb─Si alloys. The framework achieves <7% prediction error and guides the discovery of Nb38.5Ti38.5Si3Zr18V2 alloy with 22.791 MPa·m1/2 fracture toughness, breaking the 20 MPa·m1/2 barrier.
Dezhi Chen, Chao Xu, Jingyue Yu, Qi Wang, Hongze Fang, Shuo Yin, Turab Lookman, Ruirun Chen   +7 more
wiley   +1 more source

Regular Cycles of Forward and Backward Signal Propagation in Prefrontal Cortex and in Consciousness

Frontiers in Systems Neuroscience, 2016
This paper addresses two fundamental questions:(1) Is it possible to develop mathematical neural network models which can explain and replicate the way in which higher-order capabilities like intelligence, consciousness, optimization and prediction ...
Paul John Werbos, Jeffrey Jonathan (Joshua) Davis   +1 more
doaj   +1 more source

Contamination degree prediction of insulator surface based on exploratory factor analysis‐least square support vector machine combined model

High Voltage, 2021
This study presents a combined model based on the exploratory factor analysis (EFA) and the least square support vector machine (LSSVM) to predict the contamination degree of insulator surface. Firstly, EFA method is utilised to reduce numerous influence
Jiaxiang Sun, Hongru Zhang, Qingquan Li, Hongshun Liu, Xinbo Lu, Kaining Hou   +5 more
doaj   +1 more source

Discriminator‐Guided Inverse Folding for Multi‐Property Protein Design

Advanced Science, EarlyView.
Discriminator‐Guided Inverse Folding (DGIF) integrates multiple property predictors trained from single‐property datasets to guide protein sequence generation from a backbone structure. DGIF enables simultaneous improvement of thermostability and solubility without requiring multi‐property annotated datasets and generates designs that move toward the ...
Yuchuan Zheng, Chuyi Liu, Zhaoming Liu, Mao Su, Chenyu Tang, Xiang Zheng, Hao Zhang, Jingyuan Li   +7 more
wiley   +1 more source

Dual‐Mode Nanoporous SiO2 Memristors with Coexisting Volatile and Nonvolatile Dynamics for Reservoir Computing

Advanced Science, EarlyView.
A nanoporous SiO2 memristor enabling reconfigurable volatile and non‐volatile switching within a single device is demonstrated. The dual‐mode functionality supports both physical reservoir dynamics and synaptic weight storage, allowing unified hardware implementation of reservoir computing for temporal information processing, including image and ...
Bohao Ding, Tongjun Zhang, Li Shao, Nikolay Zhelev, Andrew L. Hector, Ruomeng Huang   +5 more
wiley   +1 more source

Backpropagation and F-adjoint

, 2023
This paper presents a concise mathematical framework for investigating both feed-forward and backward process, during the training to learn model weights, of an artificial neural network (ANN). Inspired from the idea of the two-step rule for backpropagation, we define a notion of F_adjoint which is aimed at a better description of the backpropagation ...
openaire   +2 more sources

Thin‐Film Transistor Based Active Taxel for Multimode Tactile Perception and Fused Processing

Advanced Science, EarlyView.
Skin serves as the largest‐area organ for human and embodied intelligent robots, providing tactile interaction. An active multimode fused (AMF) artificial skin is developed using standard TFT processes, featuring 2T‐1C taxels with optical and electrostatic capacitive receptors for cross‐modal sensing.
Sihao Wu, Zheng Zhou, Aoran Xu, Lutong Wang, Tingchen Yi, Chuanlin Sun, Haotong Zhu, Jiaqi Li, Junchen Dong, Li Zhou, Lifeng Liu, Yimao Cai, Dedong Han, Xing Zhang   +13 more
wiley   +1 more source