检索结果-内蒙古大学图书馆

Filling the Missing: Exploring Generative AI for Enhanced Federated Learning over Heterogeneous Mobile Edge Devices

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Peichun Zhang, Hanwen Wu, Yuan Qian, Liping Yu, Rong Niyato, Dusit Shen, Xuemin The State Key Laboratory of Internet of Things for Smart City University of Macau China Department of Computer and Information Science University of Macau China College of Information Engineering Zhejiang University of Technology Hangzhou310023 China School of Automation Guangdong University of Technology Guangzhou China Guangdong Key Laboratory of IoT Information Technology Guangzhou China The School of Computer Science and Engineering Nanyang Technological University Block N4-02a-32 Nanyang Avenue Singapore Singapore The Department of Electrical and Computer Engineering University of Waterloo WaterlooONN2L 3G1 Canada

Distributed Artificial Intelligence (AI) model training over mobile edge networks encounters significant challenges due to the data and resource heterogeneity of edge devices. The former hampers the convergence rate of the global model, while the latter diminishes the devices’ resource utilization efficiency. In this paper, we propose a generative AI-empowered federated learning to address these challenges by leveraging the idea of FIlling the MIssing (FIMI) portion of local data. Specifically, FIMI can be considered as a resource-aware data augmentation method that effectively mitigates the data heterogeneity while ensuring efficient FL training. We first quantify the relationship between the training data amount and the learning performance. We then study the FIMI optimization problem with the objective of minimizing the device-side overall energy consumption subject to required learning performance constraints. The decomposition-based analysis and the cross-entropy searching method are leveraged to derive the solution, where each device is assigned suitable AI-synthesized data and resource utilization policy. Experiment results demonstrate that FIMI can save up to 50% of the device-side energy to achieve the target global test accuracy in comparison with the existing methods. Meanwhile, FIMI can significantly enhance the converged global accuracy under the non-independently-and-identically distribution (non-IID) data. © 2023, CC BY.

关键词： Energy utilization

Continual Learning through Human-Robot Interaction: Human Perceptions of a Continual Learning Robot in Repeated Interactions

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Ayub, Ali De Francesco, Zachary Holthaus, Patrick Nehaniv, Chrystopher L. Dautenhahn, Kerstin Concordia University MontrealQCH3G 1M8 Canada Department of Systems Design Engineering University of Waterloo WaterlooONN2L 3G1 Canada Department of Electrical and Computer Engineering University of Waterloo WaterlooONN2L 3G1 Canada School of Physics Engineering and Computer Science University of Hertfordshire HertfordshireAL10 9AB United Kingdom

For long-term deployment in dynamic real-world environments, assistive robots must continue to learn and adapt to their environments. Researchers have developed various computational models for continual learning (CL) that can allow robots to continually learn from limited training data, and avoid forgetting previous knowledge. While these CL models can mitigate forgetting on static, systematically collected datasets, it is unclear how human users might perceive a robot that continually learns over multiple interactions with them. In this paper, © 2023, CC BY.

关键词： Adversarial machine learning

Survey on Beyond Diagonal RIS Enabled 6G Wireless Networks: Fundamentals, Recent Advances, and Challenges

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Khan, Wali Ullah Ahmed, Manzoor Sheemar, Chandan Kumar Renzo, Marco Di Lagunas, Eva Mahmood, Asad Shah, Syed Tariq Dobre, Octavia A. Querol, Jorge Chatzinotas, Symeon SnT University of Luxembourg Luxembourg City1855 Luxembourg School of Computer and Information Science Hubei Engineering University Xiaogan City432000 China School of Computer Science and Electronic Engineering University of Essex Colchester United Kingdom Faculty of Engineering and Applied Science Memorial University St. John’sNLA1C 5S7 Canada Université Paris-Saclay CNRS CentraleSupélec Lab-oratoire des Signaux et Systèmes 3 Rue Joliot-Curie Gif-sur-Yvette91192 France King’s College London Centre for Telecommunications Research Department of Engineering LondonWC2R 2LS United Kingdom

Beyond Diagonal Reconfigurable Intelligent Surfaces (BD-RIS) represent a groundbreaking innovation in sixth-generation (6G) wireless networks, enabling unprecedented control over wireless propagation environments compared to conventional diagonal RIS (D-RIS). This survey provides a comprehensive analysis of BD-RIS, detailing its architectures, operational principles, and mathematical modeling while highlighting its performance benefits. BD-RIS classifications, including single-connected, fully-connected, and group-connected architectures, and their reflective, transmissive, hybrid, and multi-sector operating modes are examined. Recent advances in BD-RIS-enabled 6G networks are reviewed, focusing on critical areas such as channel estimation, sum-rate and spectral efficiency optimization, energy efficiency enhancement, and security. The survey identifies fundamental challenges in BD-RIS research, including hardware design limitations, adaptive channel estimation, and the impact of non-ideal hardware effects. Future research directions for BD-RIS are proposed, emphasizing the integration of artificial intelligence and machine learning (AI/ML), joint optimization of communication and sensing, and enhanced physical layer security (PLS). This study concludes by underscoring BD-RIS’s transformative potential to redefine 6G wireless networks, offering valuable insights and lessons for future research and development. © 2025, CC BY.

关键词： Channel estimation

SAFFIRE: System for Autonomous Feature Filtering and Intelligent ROI Estimation 25th

学校读者我要写书评

暂无评论

SAFFIRE: System for Autonomous Feature Filtering and Intelli...

25th International Conference on Pattern Recognition Workshops, ICPR 2020

作者： Boschi, Marco Di Stefano, Luigi Alessandrini, Martino Department of Computer Science and Engineering University of Bologna Viale Risorgimento 2 Bologna40136 Italy Datalogic S.r.l. Via San Vitalino 13 Calderara di RenoBO40012 Italy

ISBN: (纸本)9783030687984

This work introduces a new framework, named SAFFIRE, to automatically extract a dominant recurrent image pattern from a set of image samples. Such a pattern shall be used to eliminate pose variations between samples, which is a common requirement in many computer vision and machine learning tasks. The framework is specialized here in the context of a machine vision system for automated product inspection. Here, it is customary to ask the user for the identification of an anchor pattern, to be used by the automated system to normalize data before further processing. Yet, this is a very sensitive operation which is intrinsically subjective and requires high expertise. Hereto, SAFFIRE provides a unique and disruptive framework for unsupervised identification of an optimal anchor pattern in a way which is fully transparent to the user. SAFFIRE is thoroughly validated on several realistic case studies for a machine vision inspection pipeline. © 2021, Springer Nature Switzerland AG.

关键词： Object detection

Scale-Adaptive viable tumor burden estimation via histopathological microscopy image segmentation

学校读者我要写书评

暂无评论

computers in Biology and Medicine 2025年 189卷 109915-109915页

作者： Sun, Yibao Xu, Zhaoyang Guo, Yihao Huang, Jian Huang, Gaopeng Huang, Tangsen Zhao, Lou Jiang, Shaowei Zheng, Zhiwen Liu, Jin Zhang, Xiaoshuai Huang, Xingru Pengcheng Laboratory Nanshan District Guangdong Shenzhen518055 China University of Cambridge Department of Paediatrics CambridgeCB2 0QQ United Kingdom School of Electronic Engineering and Computer Science Queen Mary University of London Mile End Road LondonE1 4NS United Kingdom Hangzhou Dianzi University Zhejiang Hangzhou310018 China Department of Information Science and Engineering Ocean University of China Shandong Qingdao266100 China

Cancer segmentation in whole-slide images is a fundamental step for estimating tumor burden, which is crucial for cancer assessment. However, challenges such as vague boundaries and small regions dissociated from viable tumor areas make it a complex task. Considering the usefulness of multi-scale features in various vision-related tasks, we present a structure-aware, scale-adaptive feature selection method for efficient and accurate cancer segmentation. Built on a segmentation network with a popular encoder–decoder architecture, a scale-adaptive module is proposed to select more robust features that better represent vague, non-rigid boundaries. Furthermore, a structural similarity metric is introduced to enhance tissue structure awareness and improve small region segmentation. Additionally, advanced designs, including several attention mechanisms and selective-kernel convolutions, are incorporated into the baseline network for comparative study purposes. Extensive experimental results demonstrate that the proposed structure-aware, scale-adaptive network achieves outstanding performance in liver cancer segmentation compared to the top submitted results in the PAIP 2019 challenge. Further evaluation of colorectal cancer segmentation shows that the scale-adaptive module either improves the baseline network or outperforms other advanced attention mechanism designs, particularly when considering the trade-off between efficiency and accuracy. The source code is publicly available at https://***/IMOP-lab/Scale-Adaptive-Net. © 2025

关键词： Image segmentation

Effectiveness of Image-Based Classifiers for On-Site Identification of Lichens in Reserved Forest

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Sankar, Karthikumar Rajaram, Shyam Kumar Karuppiah, Ponmurugan Periyasami, Govindasami Karthikeyan, Perumal Selvi, Chengathir M. Selvan, Basil Tamil E. Department of Computer Science and Engineering Mepco Schlenk Engineering College Tamil Nadu Sivakasi626005 India Department of Biotechnology Kamaraj College of Engineering and Technology S.P.G.C.Nagar K.Vellakulam Tamil Nadu Virudhunagar626001 India Department of Botany and Microbiology College of Science King Saud University P.O. Box 2455 Riyadh11451 Saudi Arabia 2Department of Chemistry College of Science King Saud University P.O. Box 2455 Riyadh11451 Saudi Arabia Department of Chemistry and Biochemistry Ohio State University 170A CBEC 151 Woodruff Avenue ColumbusOH43210 United States Department of Computer Science and Engineering Kamaraj College of Engineering and Technology S.P.G.C.Nagar K.Vellakulam Tamil Nadu Virudhunagar626001 India

The collection and identification of lichen samples within the reserved forest pose significant challenges, leading researchers to seek alternative methodologies. In this context, an in-depth study was conducted using two approaches to evaluate the efficacy of several network models for the identification of lichens in digital images. In the initial method, a total of 119 lichen images, comprising 15 images from experimental work and additional images from open sources, were employed for the traditional classification approach utilizing MATLAB. In which, 90 different features were extracted from various color models such as RGB, YCbCr, HSV, CMYK, LAB, YIQ and subjected for classification in ANN model, which resulted 76.6% accuracy for the class of lichen family. In another approach, each species of lichen images was augmented using generative augmentation tool and subjected for multiclass deep learning network models, precisely transfer learning model. In total, 10410 images comprising 5 thallus type, 8 order and 77 distinct lichen species were used to train three different network models (VGG16, VGG19, ***50). It was found that ***50 model exhibited highest sensitivity (>0.99) and precision. The class lichen thallus type achieved a maximum accuracy of 86.6%, while the order exhibited 80%. Overall, the lichen species under the order of Laccanorals with fruticose thallus type showed highest classification accuracy in all the three models attempted in this study. The study concluded that utilizing transfer learning with a multilayer perceptron model proved to be more effective and accurate for the lichen dataset, especially when dealing with greater complexity in picture information. The obtained results were comparable to those of other recent deep learning approaches. © 2024, The Authors. All rights reserved.

关键词： Fungi

Generating Origin-Destination Matrices in Neural Spatial Interaction Models

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zachos, Ioannis Girolami, Mark Damoulas, Theodoros Department of Engineering Cambridge University CambridgeCB2 1PZ United Kingdom The Alan Turing Institute LondonNW1 2DB United Kingdom Departments of Statistics & Computer Science University of Warwick CoventryCV4 7AL United Kingdom

Agent-based models (ABMs) are proliferating as decision-making tools across policy areas in transportation, economics, and epidemiology. In these models, a central object of interest is the discrete origin-destination matrix which captures spatial interactions and agent trip counts between locations. Existing approaches resort to continuous approximations of this matrix and subsequent ad-hoc discretisations in order to perform ABM simulation and calibration. This impedes conditioning on partially observed summary statistics, fails to explore the multimodal matrix distribution over a discrete combinatorial support, and incurs discretisation errors. To address these challenges, we introduce a computationally efficient framework that scales linearly with the number of origin-destination pairs, operates directly on the discrete combinatorial space, and learns the agents’ trip intensity through a neural differential equation that embeds spatial interactions. Our approach outperforms the prior art in terms of reconstruction error and ground truth matrix coverage, at a fraction of the computational cost. We demonstrate these benefits in large-scale spatial mobility ABMs in Cambridge, UK and Washington, DC, USA. © 2024, CC BY.

关键词： Matrix algebra

A feedforward artificial neural network model for classification and detection of type 2 diabetes 1

学校读者我要写书评

暂无评论

A feedforward artificial neural network model for classifica...

1st International Conference on Recent Trends in Applied Research, ICoRTAR 2020

作者： Frimpong, Enoch A. Oluwasanmi, Ariyo Baagyere, Edward Y. Zhiguang, Qin School of Information and Software Engineering University of Electronic Science and Technology of China No.4 North Jianshe Road Section 2 Chengdu Chengdu610054 China Department of Computer Science University for Development Studies Box TL 1350 Tamale Ghana

Efforts to enhance accuracy in medical diagnostics in molecular medicine have contributed to the wide use of artificial neural network (ANN) algorithms for disease detection due to its ability to process large medical datasets and integrate them into characterized outputs to avoid misdiagnosis. Typically, the application of ANNs have proven useful in sample analyses of patients with diabetes and in decision support systems. Over the years, various ANN models have been utilized in medical diagnostics;however, these approaches still maintain certain levels of error and have lesser training and testing accuracies in disease detection. In this study, we propose a Feedforward Artificial Neural Network (FFANN) model with a dense neural network architecture suitable for processing numeric and textual dataset. We carefully designed our model structure to have the ability to maximize the number of layers and nodes required to learn every feature of the dataset and also to perform effective computations but avoiding model under fitting and overfitting which occurs when less or more layers are used respectively. This approach puts our model ahead of other state-of-the-art prediction models which have been proposed in terms of performance as it achieved 97.27% and 96.09% training and testing accuracies, respectively, for type 2 diabetes detection on Pima Indian Diabetes dataset. © Published under licence by IOP Publishing Ltd.

关键词： Decision support systems

Radiance+: A Case Tool for Green Software Design

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Larracoechea, Jorge Andrés Ilarri, Sergio Roose, Philippe LIUPPA E2S University of Pau 2 Allée du Parc Montaury Anglet Nouvelle-Aquitaine64600 France Department of Computer Science and Systems Engineering I3A Universidad de Zaragoza C. María de Luna 3 Aragón Zaragoza50018 Spain

Regardless of the improvements in the efficiency of energy consumption of informationand communication technology, electric energy consumption will forever be a requi-site for software execution. Consequently, researchers have promoted the developmentof green and sustainable software with new development methods and tools. These,however, have been adopted with limited success, due to technicalities and specificlanguage/platform requirements. In this paper, we present RADIANCE+: a proof ofconcept web app for designing greener software with a model-driven approach basedon a Domain-Specific Modeling Language. RADIANCE+, in contrast to other tools,welcomes users with different levels of knowledge on green software and software ar-chitecture. Moreover, RADIANCE+ assesses and rates the prospective hardware andbehavioral consumption of the profiles created by the user, assisting them in identify-ing possible design changes that result in greener software from the initial stages ofsoftware development. It includes a Behavioral Load Simulator (BLS), which trans-lates the profiles into a multi-agents simulation, whose evolving prospective hardwareconsumption can be translated into the evolution of energy consumption. The resultsof the BLS can be further used by deployment simulators so that software designersand architects can experiment with different software architectures, load balancing,and scheduling or orchestration algorithms, all from the initial stages of the SoftwareDevelopment Life-Cycle. © 2024, The Authors. All rights reserved.

关键词： Energy utilization