Article
Physics, Multidisciplinary
Ginestra Bianconi
Summary: Maximum entropy network ensembles have been effective in modeling and solving inference problems in sparse network topologies. However, existing models have limitations which have been addressed by proposing hierarchical models for exchangeable networks that can handle fixed or arbitrary number of nodes in a grand canonical approach.
Article
Computer Science, Information Systems
Yetian Fan, Wenyu Yang
Summary: This paper proposes a BP algorithm with graph regularization (BPGR) to optimize the parameters and improve the generalization performance of BP neural networks. The proposed method enforces the latent features of the hidden layer to be more concentrated, enhancing the network's generalization capability. The modified graph regularization simplifies gradient calculation and better penalizes extreme weight values. Additionally, the graph regularization can be integrated with deep neural networks to further improve their generalization performance.
INFORMATION SCIENCES
(2022)
Article
Optics
Xuexiao Ma, Jiaqiang Lin, Chuansheng Dai, Jialiang Lv, Peijun Yao, Lixin Xu, Chun Gu
Summary: This paper proposes a new approach to analyze the mode-locked performance of fiber lasers using machine learning technology. By using artificial neural networks to judge pulse evolution and predict pulse shape, and combining the genetic algorithm, the authors demonstrate how to compute the parameters of fiber lasers.
OPTICS AND LASER TECHNOLOGY
(2022)
Article
Quantum Science & Technology
Chenhui Zhao, Zenan Huang, Donghui Guo
Summary: The research utilizes the SNN dynamic system model to simulate the operation mechanism and convergence of the quantum annealing algorithm, comparing the process to elastic motion in a quantum tunneling field. The results demonstrate that the SNN dynamic system model is suitable for describing the operation of the quantum annealing algorithm for optimization.
QUANTUM INFORMATION PROCESSING
(2021)
Article
Computer Science, Hardware & Architecture
Jiang Shao, Minglin Li, Xinyi Li, Guowei Liu, Sen Liu, Bin Liu, Yang Xu
Summary: Congestion control in datacenter networks aims to achieve high link utilization, low queuing delay, and rapid convergence to fairness. However, existing switch-driven schemes have either slow convergence or fail to guarantee low queues. In this paper, we propose RaceCC, a rapidly converging explicit CC scheme that achieves all three primary goals simultaneously.
JOURNAL OF NETWORK AND COMPUTER APPLICATIONS
(2023)
Article
Mathematics, Applied
Yumin Dong, Wei Liao, Mingqiu Wu, Wanbin Hu, Zhengquan Chen, Dong Hou
Summary: This study investigates the application of fractional order calculus in reducing the convergence time of neural networks. Theoretical proofs and simulations show that fractional order neural networks can achieve higher accuracy and shorter convergence time compared to integer order neural networks.
MATHEMATICAL METHODS IN THE APPLIED SCIENCES
(2022)
Article
Biochemistry & Molecular Biology
Christina G. Gangemi, Rahkesh T. T. Sabapathy, Harald Janovjak
Summary: In humans, more than 500 kinases phosphorylate approximately 15% of all proteins, forming a growing phosphorylation network. A comprehensive computational analysis revealed the existence of convergent kinase-substrate relationships (cKSRs), which are common and involve a large number of kinases and substrates. This study also demonstrated experimentally the convergence of CDK4/6 kinases in phosphorylating the tumor suppressor RB, and proposed a methodology to dissect these convergent interactions.
Article
Chemistry, Physical
Vitaly A. Gorbunov, Anastasiia I. Uliankina, Pavel Stishenko, Alexander Myshlyavtsev
Summary: The self-assembly of TPyB-Cu networks on MXene surface was investigated, revealing different local environments and thermal stabilities of metal-organic structures. The self-assembly offers opportunities for stabilizing and tuning catalyst properties.
APPLIED SURFACE SCIENCE
(2022)
Article
Automation & Control Systems
Lin Xiao, Yingkun Cao, Jianhua Dai, Lei Jia, Haiyan Tan
Summary: This article introduces a novel method for designing general zeroing neural network models with finite-time convergence (FTC) and predefined-time convergence (PTC), and proposes theoretical criteria for determining their convergence properties. These criteria can help design ZNN models with FTC or PTC more easily, and numerical experiments are conducted to verify the models with different activation functions.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2021)
Article
Computer Science, Artificial Intelligence
Hao Wang, Chi-Sing Leung, Andy Hau-Ping Chan, Anthony G. Constantinides, Wenming Cao
Summary: The aim of this research is to solve the sparse portfolio problem using continuous time neural networks, and two novel neural models are proposed to address the non-differentiable issue. Experimental results demonstrate the effectiveness of the proposed approaches over existing analog models.
Article
Automation & Control Systems
Chunhao Han, Bing Zheng, Jiao Xu
Summary: In this paper, a modified noise-tolerant zeroing neural network (MNTZNN) model is proposed. The MNTZNN model extends the NTZNN model to a more general form and significantly accelerates its convergence rate. Numerical experiments confirm the faster convergence rate of the MNTZNN model compared to the NTZNN model under certain conditions. Furthermore, the applicability and practicality of the MNTZNN model are demonstrated through its successful application to the path tracking of a 6-link planar robot manipulator under noise disturbance.
JOURNAL OF THE FRANKLIN INSTITUTE-ENGINEERING AND APPLIED MATHEMATICS
(2023)
Article
Computer Science, Artificial Intelligence
Zhe Wei, Wenwen Jia, Wei Bian, Sitian Qin
Summary: With the development of artificial intelligence and big data, distributed optimization has shown great potential in machine learning. This paper proposes a novel neural network for cooperatively solving nonsmooth distributed optimization problems, and its effectiveness and practicality are demonstrated through simulation results and a practical application.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Automation & Control Systems
Xuanyu Cao, K. J. Ray Liu
Summary: This article explores a novel generic network cost minimization problem, proposing a distributed variant of Newton's method to solve it, which demonstrates faster convergence and efficiency compared to alternative first-order optimization methods. Numerical simulations validate the effectiveness and superiority of the proposed algorithm, especially in cases of ill-conditioned cost functions. Complexity issues of the proposed distributed Newton's method and alternative first-order methods are also discussed.
IEEE TRANSACTIONS ON AUTOMATIC CONTROL
(2021)
Review
Chemistry, Multidisciplinary
Nicolai Lehnert, Bradley W. Musselman, Lance C. Seefeldt
Summary: Understanding the complex chemistry of enzymes involved in the Nitrogen Cycle is crucial in limiting anthropogenic effects on the environment.
CHEMICAL SOCIETY REVIEWS
(2021)
Article
Mathematics, Interdisciplinary Applications
Li Wang, Xingxu Chen, Juhe Sun
Summary: This paper explores the optimization problem of variational inequality with constraints, deriving the equivalent operator equations using Lagrange function and projection operator. A second-order differential equation system with controlled process is established for solving the problem, with accumulation points of the trajectory proving to be solutions. Numerical results verify the effectiveness of this approach in solving variational inequality with constraints.
