Article
Computer Science, Artificial Intelligence
Wenhao Lu, Zhengyuan Zhang, Feng Qin, Wenwen Zhang, Yuncheng Lu, Yue Liu, Yuanjin Zheng
Summary: In recent decades, there has been significant interest in the hardware implementation of feedforward neural networks. However, when implementing neural networks in analog circuits, the circuit-based model is sensitive to hardware nonidealities. This paper focuses on the presence of time-varying noise at the input of hidden neurons and proposes a noise-resilient network design to counteract its effects.
Article
Engineering, Multidisciplinary
MengYan Ge, GuoWei Wang, Ya Jia
Summary: Iterative methods were used to simulate in vitro feedforward neural networks in physiological experiments. The study investigated the effect of Gaussian colored noise and electromagnetic radiation on the propagation of subthreshold excitatory postsynaptic current signals. It was found that electromagnetic radiation slightly reduces the propagation of weak signals and an increase in feedback gain leads to longer propagation time. The feedforward neural network studied is a simple model, while more complex structures can be observed in real biological systems.
SCIENCE CHINA-TECHNOLOGICAL SCIENCES
(2021)
Article
Physics, Fluids & Plasmas
Saiya Bai, Fabing Duan, Francois Chapeau-Blondeau, Derek Abbott
Summary: Injecting noise into a neural network improves its generalization ability and expands its parameter space and weight range. This configuration constitutes a stochastic-resonance-based threshold neural network, and the injected noise acts as a regularizer.
Article
Engineering, Mechanical
Hao Si, Xiaojuan Sun
Summary: This paper investigates the propagation of population firing rate and pulse packets in the cortical neural network. The results show that proper feedforward connection probability and strength can promote the propagation of population firing rate, while increasing the feedforward synaptic time constant can enhance the fidelity of the propagating population firing rate. Adjusting the relative strength, recurrent probability, and feedforward synaptic connection can also promote the propagation of pulse packets, with a larger feedforward synaptic time constant having different effects on population firing rate and pulse packets propagation.
NONLINEAR DYNAMICS
(2021)
Article
Chemistry, Multidisciplinary
Dongshin Kim, Jang-Sik Lee
Summary: Neurotransmitters play a crucial role in controlling signal transmission in the nervous system, and the balance between excitatory and inhibitory synaptic responses is fundamental for the characteristics of the nervous system. This study develops artificial synapses that emulate the excitatory and inhibitory functions of biological synapses, paving the way for bio-realistic neuromorphic devices.
ADVANCED FUNCTIONAL MATERIALS
(2022)
Article
Remote Sensing
Mingchen Zhu, Xianwen Yu, Wei Sun
Summary: This article introduces a high temporal and high spatial resolution weighted mean temperature (T-m) model for the region of China. The model, named CTm-FNN, combines the data from the European Centre for Medium-Range Weather Forecasts with the feedforward neural network algorithm to achieve high accuracy. The CTm-FNN model shows significant improvement in terms of error reduction and standard deviation compared to other models.
Article
Computer Science, Artificial Intelligence
Pui-Wai Ma, T. -H. Hubert Chan
Summary: We propose a new type of feedforward neural network that is equivariant with respect to the unitary group U(n). The network can handle input and output vectors in Cn with arbitrary dimension n, without the need for convolution layers. Our implementation avoids errors caused by truncated higher order terms in Fourier-like transformations, and each layer can be efficiently implemented using simple calculations. As a proof of concept, we provide empirical results on predicting atomic motion dynamics, demonstrating the practicality of our approach.
Article
Computer Science, Information Systems
Adriano Mourthe, Carlos E. Mello
Summary: Collaborative Filtering has been extensively studied, and neural-based methods have achieved great success in personalized recommendation. However, the complexity of these methods increases the computational cost, and reducing the sparsity and dimensionality of input features is crucial for improving accuracy.
INFORMATION SCIENCES
(2022)
Article
Physiology
Xiaoxiao Peng, Wei Lin
Summary: This study proposes a model of noise-perturbed random neural networks with both excitatory and inhibitory populations. By using mean-field theory, the researchers obtain an equivalent system and investigate the stationary autocorrelation functions to analyze the synchronized behaviors and chaotic dynamics of the two populations. The results show that noise can suppress chaos, while correlation coefficient in intra-coupling strengths can enhance chaos occurrence. The study also detects a phenomenon where the parameters region corresponds to neither linearly stable nor chaotic dynamics.
FRONTIERS IN PHYSIOLOGY
(2022)
Article
Engineering, Mechanical
Shuli Wei, Jian Wang, Jinping Ou
Summary: The study focuses on enhancing the accuracy of MR damper modeling using a neural network approach, proposing the use of Hilbert transformation to construct instantaneous variables and an indirect modeling method. Comparative analysis shows significant improvements in predicting damper force nonlinearity.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2021)
Article
Engineering, Environmental
Ankun Xu, Huimin Chang, Yingjie Xu, Rong Li, Xiang Li, Yan Zhao
Summary: The research reviewed the application of artificial neural networks in different scales of waste management, finding that they are widely employed in waste generation prediction and technological parameter estimation. Most studies included a data size of 101-150 and optimal numbers of hidden layer nodes range from 4 to 20. The review aims to provide basic and comprehensive knowledge for researchers in general waste management and specialized ANN study on solid waste-related issues.
Article
Mathematics, Applied
Hsin-Chieh Wu, Tin-Chih Toly Chen, Min-Chi Chiu
Summary: This study explores how to fuzzify a feedforward neural network to generate fuzzy forecasts containing actual values while minimizing range; by independently fuzzifying parameters to theoretically derive optimal values; and found that fuzzifying the output node's threshold and connection weights is more likely to achieve a 100% hit rate.
Article
Computer Science, Artificial Intelligence
Shaokai Zhao, Bin Chen, Hui Wang, Zhiyuan Luo, Tao Zhang
Summary: A novel feed-forward neural network inspired by the structure of the dentate gyrus and neural oscillatory analysis has been proposed to increase the storage capacity of Hopfield network. By using a mouse model of environmental enrichment and neural oscillatory analysis, a computational model of the dentate gyrus associated with better pattern separation ability has been obtained. The simulation results show significant expansion in storage capacity and improved pattern separation compared to the standard Hopfield network.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Yejiang Yang, Tao Wang, Jefferson P. Woolard, Weiming Xiang
Summary: This paper discusses the key measure of approximation error in model validation and verification for neural networks, proposing a concept of guaranteed error estimation and developing methods to efficiently compute upper-bounds of approximation errors. Through examples, the effectiveness of the approach is demonstrated.
Review
Computer Science, Artificial Intelligence
Weiwei Jiang
Summary: This article provides a latest review of deep learning models for stock market prediction, categorizing data sources, neural network structures, and evaluation metrics to help researchers stay updated and easily reproduce previous studies. It also highlights some future research directions in this topic.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Mathematical & Computational Biology
Gang Wang, Ming Yi, Sanyi Tang
Summary: In this paper, an antitumour model is proposed to characterize the processes of radiotherapy and immunotherapy. The stability and effectiveness of the model are proved through mathematical analysis. Numerical simulations indicate that radioimmunotherapy is more effective than radiotherapy or immunotherapy alone.
COMPUTATIONAL AND MATHEMATICAL METHODS IN MEDICINE
(2022)
Article
Biochemistry & Molecular Biology
Jing Zhang, Yudan Yang, Zejing Mao, Qingqing Yan, Qi Chen, Ming Yi, Yanchun Shao
Summary: In this study, it was found that MrSir2 protein has impacts on the development and MonAzPs production of Monascus ruber. MrSir2 promotes the transition from the primary growth phase to mycelial aging through regulating gene expression involved in macromolecular metabolism and cell wall synthesis. Additionally, the Delta mrsir2 strain exhibited accelerated mycelial aging, increased spore production, enhanced resistance to oxidative stress, and higher MonAzPs production.
APPLIED BIOCHEMISTRY AND BIOTECHNOLOGY
(2022)
Article
Biotechnology & Applied Microbiology
Yuan Zhu, Houwang Zhang, Yuanhang Yang, Chaoyang Zhang, Le Ou-Yang, Litai Bai, Minghua Deng, Ming Yi, Song Liu, Chao Wang
Summary: This study successfully identified potential pan-cancer-related genes by establishing a pan-cancer network and using integrative network analysis, demonstrating the potential of the method for further biological experimental verification.
BRIEFINGS IN FUNCTIONAL GENOMICS
(2022)
Article
Engineering, Multidisciplinary
LuLu Lu, Ming Yi, XiaoQian Liu
Summary: The firing sequence of the neuron system consumes a significant amount of energy, but when system parameters are tuned, the neurons can exhibit complex bursting kinetics, with the period-n bursting state carrying high amounts of information while consuming less energy per unit of information, resulting in higher energy efficiency. The mixed discharge state, where the neuron is in several bursting states simultaneously, is more energy efficient, and appropriate electromagnetic induction can enhance the neuron's energy efficiency. Optimal system parameters exist that maximize the energy efficiency of firing modes, demonstrating that the neuron carries high amounts of information while consuming less energy per unit of information.
SCIENCE CHINA-TECHNOLOGICAL SCIENCES
(2022)
Article
Physics, Multidisciplinary
Qian Cheng, Yuangen Yao, Min Li, Zhouchao Wei, Ming Yi
Summary: This paper investigates the impact of memory effects on the logic operation of the Set-Reset latch and presents experimental results using harmonic-driven operation in a nonlinear fractional-order two-well potential system. The study reveals that certain harmonic driving forces can eliminate the memory effects, enabling the system to form a stable logic gate and implement logical vibrational resonance (LVR).
EUROPEAN PHYSICAL JOURNAL PLUS
(2022)
Article
Mathematics, Applied
Lulu Lu, Zhuoheng Gao, Zhouchao Wei, Ming Yi
Summary: The study investigates the effects of excitatory-inhibitory balance and neural network structures on working memory tasks using a neuron-astrocyte network model. The results reveal that performance metrics are higher for scale-free networks compared to other structures, and the tasks can be successfully completed when the proportion of excitatory neurons exceeds 30%. An optimal region is identified for excitatory neuron proportion and synaptic weight, where memory performance metrics are higher. The study also highlights the overlap of spatial calcium patterns in the astrocyte network for different items in multi-item working memory tasks, suggesting similarities to cognitive memory formation in the brain. Additionally, cued recall in complex image tasks reduces systematic noise and maintains task stability.
Article
Chemistry, Physical
Tao Lin, Jiacheng Lin, Xiaoyao Wei, Lulu Lu, Xuefeng Yin
Summary: Among the electrode materials of supercapacitors, transition metal oxides, including MnO2, have shown promising properties. However, the low conductivity, agglomeration, and volume change of MnO2 lead to its limited specific capacitance. To address this issue, attapulgite is introduced as a composite material to enhance the electrochemical performance. The ATP-MnO2 composites exhibit higher specific capacitance and better cycle stability compared to pure MnO2 nanoflowers.
INTERNATIONAL JOURNAL OF HYDROGEN ENERGY
(2023)
Article
Neurosciences
Xiaoqian Liu, Lulu Lu, Yuan Zhu, Ming Yi
Summary: This article simulates the up and down transitions of membrane potentials using a neural network and discusses the energy consumption and efficiency. The study finds that the dynamics of intrinsic currents have a significant impact on energy consumption and efficiency. Adjusting network parameters can reduce energy consumption and enhance energy efficiency, providing insights into the metabolic consumption of spontaneous activities.
JOURNAL OF NEUROPHYSIOLOGY
(2023)
Article
Physics, Multidisciplinary
Huixia Liu, Lulu Lu, Yuan Zhu, Zhouchao Wei, Ming Yi
Summary: We investigated the response of neural networks with different topologies to weak signals and analyzed the effects of noise and electromagnetic induction on stochastic resonance. The results showed that appropriate noise and electromagnetic induction can promote the generation of stochastic resonance, and different networks exhibit different responses to stochastic resonance.
PHYSICA A-STATISTICAL MECHANICS AND ITS APPLICATIONS
(2022)
Article
Psychiatry
Lulu Lu, Hongxian Shen, Liwen Tan, Qiuping Huang, Qiongni Chen, Mining Liang, Li He, Yang Zhou
Summary: This study aimed to determine the prevalence of anxiety and depression among community-dwelling older adults in China and explore the associated factors. The results showed a high prevalence of anxiety (32.74%) and depression (37.34%) among the study participants. Gender, employment status, physical activity, physical pain, comorbidities, and social support were found to be associated with psychological health problems in older adults.
Article
Mathematics, Interdisciplinary Applications
Bo Li, Tian Huang
Summary: This paper proposes an approximate optimal strategy based on a piecewise parameterization and optimization (PPAO) method for solving optimization problems in stochastic control systems. The method obtains a piecewise parameter control by solving first-order differential equations, which simplifies the control form and ensures a small model error.
CHAOS SOLITONS & FRACTALS
(2024)
Article
Mathematics, Interdisciplinary Applications
Guram Mikaberidze, Sayantan Nag Chowdhury, Alan Hastings, Raissa M. D'Souza
Summary: This study explores the collective behavior of interacting entities, focusing on the co-evolution of diverse mobile agents in a heterogeneous environment network. Increasing agent density, introducing heterogeneity, and designing the network structure intelligently can promote agent cohesion.
CHAOS SOLITONS & FRACTALS
(2024)
Article
Mathematics, Interdisciplinary Applications
Gengxiang Wang, Yang Liu, Caishan Liu
Summary: This investigation studies the impact behavior of a contact body in a fluidic environment. A dissipated coefficient is introduced to describe the energy dissipation caused by hydrodynamic forces. A new fluid damping factor is derived to depict the coupling between liquid and solid, as well as the coupling between solid and solid. A new coefficient of restitution (CoR) is proposed to determine the actual physical impact. A new contact force model with a fluid damping factor tailored for immersed collision events is proposed.
CHAOS SOLITONS & FRACTALS
(2024)