Article
Environmental Sciences
Jiarui Zhang, Zhiyi Fu, Yilin Zhu, Bin Wang, Keran Sun, Feng Zhang
Summary: Land cover mapping is crucial for global resource monitoring, sustainable development research, and effective management. However, the complexity and computational requirements often cause delays in data processing and product publication, posing challenges for creating large-area products for monitoring dynamic land cover. This study proposes the HALF framework to automate and improve the efficiency of land cover mapping processes.
Article
Engineering, Multidisciplinary
Govind Saraswat, Vivek Khatana, Sourav Patel, Murti V. Salapaka
Summary: This article presents a finite-time stopping criterion for consensus algorithms in networks with dynamic communication topology. It proposes a maximum-minimum protocol to propagate global maximum and minimum values of agent states, allowing for a distributive determination of convergence in finite time. Experimental results illustrate the practical utility of the algorithm.
IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING
(2023)
Article
Computer Science, Information Systems
Tayebeh Bahreini, Daniel Grosu
Summary: This article addresses the Multi-Component Application Placement Problem in Mobile Edge Computing systems. It formulates the problem as a Mixed Integer Non-Linear Program and designs two algorithms for solving it. Experimental results show that the proposed algorithms perform well in terms of both execution time and solution quality.
IEEE TRANSACTIONS ON CLOUD COMPUTING
(2022)
Article
Computer Science, Hardware & Architecture
Xin-miao Chen, Shi Wang, Yong-jin Ye, Yong-zheng Wu, Bo Jiang
Summary: This paper introduces a strategy based on the business of each individual qubit to improve the execution efficiency of quantum circuits by inserting SWAP gates. The strategy exploits the uneven distribution of quantum gates over qubits to reduce the time overhead caused by inserted SWAP gates, while minimizing the negative impact on subsequent operations.
JOURNAL OF SUPERCOMPUTING
(2023)
Article
Computer Science, Information Systems
Amir Iranmanesh, Hamid Reza Naji
Summary: This study presents an approach to workflow task scheduling based on genetic algorithms, which utilizes new genetic operators and load balancing routines to enhance efficiency in cloud environments. Results demonstrate that the proposed algorithm outperforms state-of-the-art methods in task scheduling, achieving shorter makespan and lower cost.
CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS
(2021)
Article
Computer Science, Hardware & Architecture
Francesco Malandrino, Carla Fabiana Chiasserini, Nuria Molner, Antonio de la Oliva
Summary: This paper investigates the impact of network topology in distributed machine learning, presenting a system model and determining the cooperation of learning and information nodes as well as the number of epochs to minimize learning cost and meet performance requirements. Experimental results show that the proposed algorithm, DoubleClimb, outperforms state-of-the-art alternatives.
IEEE-ACM TRANSACTIONS ON NETWORKING
(2023)
Review
Green & Sustainable Science & Technology
Rudai Shan, Lars Junghans
Summary: This study presents a systematic review of optimization methods for building facade optimization (BFO), comparing the efficiency and effectiveness of different algorithms. Key findings highlight the robust feasibility and effectiveness of optimization algorithms, methods, and techniques in resolving a diverse range of BFO challenges.
Article
Computer Science, Artificial Intelligence
Christopher Renkavieski, Rafael Stubs Parpinelli
Summary: Truss optimization is an important engineering problem that can reduce material use and construction costs. Meta-heuristic algorithms, like EB-A-SHADE, have been applied to this problem due to their stochastic nature. The systematic mapping and experimental results show that EB-A-SHADE is competitive and stable for truss size optimization.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Information Systems
Gil Einziger, Gabriel Scalosub, Carla Fabiana Chiasserini, Francesco Malandrino
Summary: Efficiently deploying services while meeting quality requirements is a major challenge in network slicing. This study proposes an algorithm called REShare that can adapt to operational conditions and find an optimal balance between conflicting requirements, improving service efficiency and reducing costs.
IEEE TRANSACTIONS ON CLOUD COMPUTING
(2023)
Article
Multidisciplinary Sciences
Xiaodong Weng, Yi Liu, Changqing Xu, Xiaoling Lin, Linjun Zhan, Shunyao Wang, Dongdong Chen, Yintang Yang
Summary: Network on chip (NoC) is a promising solution to multi-core System-on-Chip (SoC) communication design challenges. However, the mapping problem in NoC is confirmed as an NP-hard problem, and existing heuristic algorithms may lead to suboptimal solutions. In this paper, a machine learning mapping algorithm is proposed, which achieves high model accuracy and mapping accuracy.
Article
Computer Science, Information Systems
Qing-Wei Chai, Wei-Min Zheng, Lili Xu, Lyuchao Liao
Summary: Optimization problems are common and challenging to solve. In this study, a novel method for denoising ECG signals is proposed, which combines a heuristic algorithm with an adaptive filtering algorithm to adjust the weight parameters of the filter. A new heuristic algorithm called CAFMO is also introduced, which incorporates a chaotic strategy into the AFMO algorithm. The CAFMO algorithm demonstrates superior performance in noise mitigation compared to other algorithms, resulting in a significant improvement in the denoising method.
Article
Computer Science, Information Systems
Nilayam Kumar Kamila, Jaroslav Frnda, Subhendu Kumar Pani, Rashmi Das, Sardar M. N. Islam, P. K. Bharti, Kamalakanta Muduli
Summary: This paper proposes integrating the concept of high-performance computing with artificial intelligence machine learning techniques in cloud platforms. The networking and computing performance data are used to validate, predict, and classify traffic and performance patterns to ensure system performance and continuous traffic flow resiliency decisions. The simulation results show that the machine learning integrated design performs 38.15% faster in traffic resilience and saves 7.5% of business costs compared to existing non-machine learning design models.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Computer Science, Information Systems
Ankita Srivastava, Narander Kumar
Summary: This article proposes an efficient load balancing algorithm based on the Firefly and Honeybee algorithm to optimize resource utilization, average load, and response time. The simulation results show improved performance in terms of average load, response time, resource usage, and imbalance degree.
CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Hardware & Architecture
Ji Liu, Xiao Liang, Wenxi Ruan, Bo Zhang
Summary: The aim of the research is to improve the efficiency of medical data processing and establish a sound medical data management system using technologies such as the PRF algorithm, ADDPC method, and BPT-CNN model. The results show that these technologies applied on a distributed computing platform can effectively enhance algorithm performance, optimize data communication cost and efficiency, and adapt to different types of data sets to prevent data loss and fragmentation.
JOURNAL OF SUPERCOMPUTING
(2022)
Article
Computer Science, Information Systems
B. Mohammad Hasani Zade, M. M. Javidi, N. Mansouri
Summary: This paper proposes two approaches to provide a secure connection between users and servers and handle large and medium task size problems. The first approach is a multi-objective scheduling algorithm based on the New Caledonian Crow Learning Algorithm (NCCLA), which aims to minimize workflow makespan and cost while increasing the cost of attack from invaders. The second approach modifies the structure of virtual machines to increase the cost of attack. Experimental results show that the proposed algorithm outperforms existing workflow algorithms in terms of FS-metric.
PEER-TO-PEER NETWORKING AND APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Qian Wan, Luona Wei, Xinhai Chen, Jie Liu
Summary: This paper proposes a Region-based Hypergraph Network (RHGN) for joint entity and relation extraction, which introduces the concept of regional hypernodes and constructs a region-based relation hypergraph to improve performance. Experimental results demonstrate the superior performance of the model in both entity recognition and relation extraction.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Hardware & Architecture
Rui Xu, Sheng Ma, Yaohua Wang, Xinhai Chen, Yang Guo
Summary: The systolic array architecture is popular for convolutional neural network hardware accelerators due to its simple and efficient design. However, challenges arise when processing special types of convolution, leading to decreased PE utilization. To address this, a configurable multi-directional systolic array (CMSA) was designed with improved data paths and PE units to increase utilization rates.
ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION
(2021)
Article
Computer Science, Software Engineering
Xinhai Chen, Jie Liu, Chunye Gong, Shengguo Li, Yufei Pang, Bo Chen
Summary: The quality control in CFD pre-processing aims to indicate the validity of the generated mesh to engineers. Existing quality measures mainly focus on subjective evaluation of shape information, neglecting mesh distribution details, leading to manual re-evaluation that increases costs and hinders automation. This study proposes an automatic 3-D structured mesh validity evaluation framework using deep neural networks, showcasing potential for deep neural networks in 3-D mesh validity evaluation and the effectiveness of MVE-Net.
COMPUTER-AIDED DESIGN
(2021)
Article
Multidisciplinary Sciences
Xinhai Chen, Rongliang Chen, Qian Wan, Rui Xu, Jie Liu
Summary: This paper introduces an improved data-free surrogate model, DFS-Net, based on deep neural networks, which uses a weighting mechanism and neural network structure to address the instability or inaccuracy in predicting PDEs. Experimental results demonstrate that DFS-Net achieves a good trade-off between accuracy and efficiency.
SCIENTIFIC REPORTS
(2021)
Article
Computer Science, Interdisciplinary Applications
Xinhai Chen, Tiejun Li, Qian Wan, Xiaoyu He, Chunye Gong, Yufei Pang, Jie Liu
Summary: In this paper, a novel differential method called MGNet is introduced for structured mesh generation. The method poses meshing as an optimization problem and utilizes a well-designed neural network to learn potential meshing rules. Experimental results show that MGNet is capable of generating acceptable meshes with desired number of cells and performs comparably or superiorly to traditional methods and other neural network-based solvers.
ENGINEERING WITH COMPUTERS
(2022)
Article
Computer Science, Interdisciplinary Applications
Qingyang Zhang, Xiaowei Guo, Xinhai Chen, Chuanfu Xu, Jie Liu
Summary: Physics-informed neural networks (PINNs) have emerged as a new approach to solve partial differential equations. This paper introduces a novel neural network structure, PINN-FFHT, for fluid flow and heat transfer problems. PINN-FFHT can predict the flow field and consider the influence of flow on the temperature field to solve the energy equation. It also proposes a flexible boundary condition enforcement method and a dynamic strategy to balance the loss term of velocity and temperature for training PINN-FFHT, resulting in faster convergence and higher accuracy compared to traditional PINN methods.
INTERNATIONAL JOURNAL OF MODERN PHYSICS C
(2022)
Article
Computer Science, Theory & Methods
Xinhai Chen, Chunye Gong, Jie Liu, Yufei Pang, Liang Deng, Lihua Chi, Kenli Li
Summary: Evaluating mesh quality before solving is crucial for error control in airfoil numerical simulation. Traditional metrics fail to recognize numerical errors caused by mesh density or distribution. Therefore, this paper introduces deep neural networks to improve automation and efficiency in airfoil mesh quality evaluation.
JOURNAL OF PARALLEL AND DISTRIBUTED COMPUTING
(2022)
Article
Engineering, Multidisciplinary
Liang Deng, Jianqiang Chen, Yueqing Wang, Xinhai Chen, Fang Wang, Jie Liu
Summary: This paper proposes a deep learning model for vortex detection using weak supervision approach. It utilizes an automatic clustering method and multi-view learning strategy to improve efficiency and accuracy. A physics-informed loss function is introduced to consider the characteristics of flow fields.
ENGINEERING APPLICATIONS OF COMPUTATIONAL FLUID MECHANICS
(2022)
Article
Computer Science, Interdisciplinary Applications
Zhichao Wang, Xinhai Chen, Tieju Li, Chunye Gong, Yufei Pang, Jie Liu
Summary: The quality of the finite element mesh has a significant impact on the efficiency and accuracy of computational fluid dynamics simulations. However, there is a lack of justifiable thresholds for determining the quality of the generated mesh. This paper proposes a novel approach to evaluate mesh quality as a graph classification problem using deep graph neural networks.
ENGINEERING WITH COMPUTERS
(2022)
Article
Physics, Multidisciplinary
Xinhai Chen, Zhichao Wang, Jie Liu, Chunye Gong, Yufei Pang
Summary: Evaluating mesh quality is crucial for accurate cylinder modeling in computational fluid dynamics (CFD) simulations. Traditional indicators are often insufficient and require manual re-evaluation. This paper presents an efficient quality indicator, Mesh-Net, that can automatically evaluate mesh quality without manual interactions.
Article
Mechanics
Xinhai Chen, Jie Liu, Qingyang Zhang, Jianpeng Liu, Qinglin Wang, Liang Deng, Yufei Pang
Summary: In this paper, a novel structured mesh generation method called MeshNet is proposed, which uses deep neural networks to learn high-quality meshing rules and generate desired meshes. The method employs a physics-informed neural network to approximate the transformation between computational and physical domains, and is trained using differential equations, boundary conditions, and a priori data. The results of experiments show that MeshNet is fast and robust, outperforming state-of-the-art neural network-based generators and traditional meshing methods.
Proceedings Paper
Computer Science, Artificial Intelligence
Chunye Gong, Xinhai Chen, Shuling Lv, Jie Liu, Bo Yang, QingLin Wang, Weimin Bao, Yufei Pang, Yang Sun
Summary: In this paper, an efficient im2col algorithm named im2cole is introduced for CNN, handling different stride and pad conditions and improving overall performance.
2021 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN)
(2021)
Article
Engineering, Mechanical
Xinhai Chen, Chunye Gong, Qian Wan, Liang Deng, Yunbo Wan, Yang Liu, Bo Chen, Jie Liu
Summary: This study improves the accuracy of predicted solutions and achieves a 97.3% performance boost on widely used surrogate models by applying transfer learning to DNN-based PDE solving tasks.
ADVANCES IN AERODYNAMICS
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Xinhai Chen, Jie Liu, Chunye Gong, Yufei Pang, Bo Chen
2020 12TH INTERNATIONAL CONFERENCE ON ADVANCED COMPUTATIONAL INTELLIGENCE (ICACI)
(2020)