Article
Computer Science, Artificial Intelligence
Weipeng Cao, Zhongwu Xie, Jianqiang Li, Zhiwu Xu, Zhong Ming, Xizhao Wang
Summary: A novel bidirectional SCN algorithm (BSCN) is proposed to accelerate training speed and improve the quality of hidden nodes, achieving faster speed, higher stability, and better generalization ability than SCN in various experiments.
Article
Computer Science, Artificial Intelligence
Hang Yu, Jie Lu, Guangquan Zhang
Summary: This article proposes a novel evolving-fuzzy-neuro system called TLFRNN, which self-organizes each layer using online topology learning algorithm and learns multiple fuzzy sets to reduce the impact of noise. By considering both fuzzy and random information in a simple inference, TLFRNN can easily detect and adapt to concept drift.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2022)
Article
Chemistry, Analytical
George Papastergiou, Apostolos Xenakis, Costas Chaikalis, Dimitrios Kosmanos, Periklis Chatzimisios, Nicholas S. Samaras
Summary: In dense deployments of wireless sensor networks, sensor placement, coverage, connectivity, and energy constraints are crucial for network lifetime. However, maintaining a balance among these conflicting constraints is challenging in large-scale networks. This paper formulates a problem of topology control and lifetime extension through sensor placement, under coverage and energy constraints, and solves it using various neural network configurations. Simulation results show that the proposed algorithm improves network lifetime while meeting communication and energy constraints in medium- and large-scale deployments.
Article
Computer Science, Artificial Intelligence
Alexandros Goulas, Fabrizio Damicelli, Claus C. Hilgetag
Summary: Biological neuronal networks serve as inspiration for artificial neuronal networks, but they are sculpted by evolution while ANNs are engineered for specific tasks. The network topology of these systems shows pronounced differences, with strategies explored to construct bio-instantiated RNNs for different species' brains. Performance of these RNNs in working memory tasks is examined, highlighting the importance of empirical data for constructing neural networks.
Article
Mathematics
Massimiliano Turchetto, Michele Bellingeri, Roberto Alfieri, Ngoc-Kim-Khanh Nguyen, Quang Nguyen, Davide Cassi
Summary: Investigating the network response to node removal and the efficacy of the node removal strategies is fundamental to network science. In this study, we propose four new measures of node centrality based on random walk and compare them with existing strategies for synthesizing and real-world networks. The results indicate that the degree nodes attack is the best strategy overall, and the new node removal strategies based on random walk show the highest efficacy in relation to specific network topology.
Article
Biochemical Research Methods
Kevin Chow, Aisharjya Sarkar, Rasha Elhesha, Pietro Cinaglia, Ahmet Ay, Tamer Kahveci
Summary: The study presents a novel alignment based network construction algorithm ANCA, which accurately predicts missing target networks, scales to large-scale biological networks and successfully identifies key temporal changes in biological networks. The method can be used to discover important genes and temporal functional changes in biological networks by focusing on topological differences.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2021)
Article
Green & Sustainable Science & Technology
Ge Chen, Hongcai Zhang, Hongxun Hui, Ningyi Dai, Yonghua Song
Summary: Thermostatically controlled loads (TCLs) are a desirable source of demand-side flexibility in distribution networks, but require consideration of power flow constraints to avoid violations. This paper proposes a novel learning-based optimal power flow (OPF) method that uses MLPs trained on historical operation data to optimize TCLs for regulation services, achieving better power scheduling performance compared to existing models. The method converts MLPs into linear constraints with binary variables to effectively solve the optimization problem, providing feasibility and optimality in TCL power scheduling.
IEEE TRANSACTIONS ON SUSTAINABLE ENERGY
(2021)
Article
Computer Science, Artificial Intelligence
Arend Hintze, Christoph Adami
Summary: Artificial neural networks (ANNs) are a promising tool for developing general artificial intelligence. This study investigates the impact of training methods on information propagation in the brain. The research finds that neuroevolution training leads to more focused information transfer in small groups of neurons and greater robustness to weight perturbations compared to backpropagation training.
NEURAL COMPUTING & APPLICATIONS
(2022)
Article
Biochemistry & Molecular Biology
Angelica M. Walker, Ashley Cliff, Jonathon Romero, Manesh B. Shah, Piet Jones, Joao Gabriel Felipe Machado Gazolla, Daniel A. Jacobson, David Kainer
Summary: Gene-to-gene networks are important tools for studying relationships between genes. Random Forest and iterative Random Forest methods can produce high-quality gene-to-gene networks. The study validates the use of synthetic and empirical data to compare these methods.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2022)
Article
Engineering, Electrical & Electronic
Yonghan Wu, Min Zhang, Lifang Zhang, Jin Li, Xue Chen, Danshi Wang
Summary: Digital twin (DT) is a promising technology that bridges the gap between physical and digital space, enabling interactive functions and intelligent control in optical networks. This paper proposes a multifactor-associated network topology portrait (NTP) scheme that provides a dynamic and comprehensive representation of the DT optical network (DTON). Multiple parameters from nodes and links are jointly evaluated, considering both network and physical layer characteristics. The performance of dynamic routing computation using six different routing algorithms is studied based on the NTP, sacrificing a little bit of routing computation time for significant improvements in outage probability, outage latency, and the number of service requests.
JOURNAL OF LIGHTWAVE TECHNOLOGY
(2023)
Article
Automation & Control Systems
Housheng Su, Dan Chen, Gui-Jun Pan, Zhigang Zeng
Summary: The article discusses the dependency of indicators constructed by the network adjacency matrix and Laplacian matrix on the topological structure of the network, and proves from various aspects that spectral entropy has a better ability to identify global topology variations compared to traditional distribution entropy.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Energy & Fuels
Haotian Ge, Bingyin Xu, Xinhui Zhang, Yongjian Bi, Zida Zhao
Summary: This paper addresses the lack of feeder topology information exchange mechanism in IEC 61850 by developing an information model, and proposes a mechanism for topology information exchange between IEDs. The effectiveness of the method is verified through experiments, demonstrating the potential for testing the validity and interoperability of the proposed model with more distributed applications.
Article
Computer Science, Artificial Intelligence
Enzo Tartaglione, Andrea Bragagnolo, Francesco Odierna, Attilio Fiandrotti, Marco Grangetto
Summary: The method utilizes neuron sensitivity as a regularization term to learn sparse topologies and effectively compress neural networks for resource-constrained devices. Experimental results show that the method achieves competitive compression ratios compared to state-of-the-art references.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Engineering, Electrical & Electronic
Fangzhou Wu, Li Chen, Nan Zhao, Yunfei Chen, F. Richard Yu, Guo Wei
Summary: This study proposes a new computing technique for data aggregation in multi-hop wireless networks. By combining CoMAC and orthogonal communication, it is possible to compute functions more efficiently and improve network performance.
IEEE TRANSACTIONS ON COMMUNICATIONS
(2021)
Article
Computer Science, Hardware & Architecture
Yuhang Zhou, Zhiqun Gu, Jiawei Zhang, Yuefeng Ji
Summary: Using NFPs as aerial relays is a promising solution for mobile backhaul frameworks in 5G and beyond wireless networks. FSO technology has become an alternative to RF technology for backhaul links due to its advantages. However, the fragility of FSO links and the limited number of transceivers carried by each NFP pose challenges in deploying NFPs to meet network traffic demands. To address these challenges, a joint optimization scheme of NFP deployment and network topology formation is proposed, which outperforms the single optimization scheme. A heuristic algorithm is also developed to efficiently solve the joint optimization problem in large-scale network scenarios.
JOURNAL OF OPTICAL COMMUNICATIONS AND NETWORKING
(2023)