检索结果-内蒙古大学图书馆

Exploring Bidirectional Bounds for Minimax-Training of Energy-Based Models

International Journal of computer Vision 2025年 1-22页

作者： Geng, Cong Wang, Jia Chen, Li Gao, Zhiyong Frellsen, Jes Hauberg, Søren China Mobile Research Institute Beijing 100032 China Institute of Image Communication and Network Engineering Shanghai Jiao Tong University Shanghai 200240 China Department of Applied Mathematics and Computer Science Technical University of Denmark Richard Petersens Plads Kgs.@ Lyngby 2800 Denmark

Energy-based models (EBMs) estimate unnormalized densities in an elegant framework, but they are generally difficult to train. Recent work has linked EBMs to generative adversarial networks, by noting that they can be trained through a minimax game using a variational lower bound. To avoid the instabilities caused by minimizing a lower bound, we propose to instead work with bidirectional bounds, meaning that we maximize a lower bound and minimize an upper bound when training the EBM. We investigate four different bounds on the log-likelihood derived from different perspectives. We derive lower bounds based on the singular values of the generator Jacobian and on mutual information. To upper bound the negative log-likelihood, we consider a gradient penalty-like bound, as well as one based on diffusion processes. In all cases, we provide algorithms for evaluating the bounds. We compare the different bounds to investigate, the pros and cons of the different approaches. Finally, we demonstrate that the use of bidirectional bounds stabilizes EBM training and yields high-quality density estimation and sample generation. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025.

关键词： Bidirectional bounds Density estimation Energy-based model Minimax-training

来源：评论

学校读者我要写书评

暂无评论

A Deep Multi-Task Learning Network for Activity Recognition and User Identification Using Smartphone Sensors 16

A Deep Multi-Task Learning Network for Activity Recognition ...

引用

16th International Conference on ENTERprise Information Systems, CENTERIS 2024 - 12th International Conference on Project MANagement, ProjMAN 2024 - 14th International Conference on Health and Social Care Information Systems and Technologies, HCist 2024

作者： Mekruksavanich, Sakorn Phaphan, Wikanda Jitpattanakul, Anuchit Department of Computer Engineering School of Information and Communication Technology University of Phayao Phayao56000 Thailand Research Group in Statistical Learning and Inference Department of Applied Statistics Faculty of Applied Science King Mongkut's University of Technology North Bangkok Bangkok10800 Thailand Intelligent and Nonlinear Dynamic Innovations Research Center Department of Mathematics Faculty of Applied Science King Mongkut's University of Technology North Bangkok Bangkok10800 Thailand

Human activity recognition (HAR) and user identification utilizing smartphone sensors are crucial in domains such as wellness tracking, tailored services, and security. Conventional approaches handle these activities as distinct entities, leading to less-Than-ideal performance and restricted applicability. This research presents a novel deep multi-Task learning network that simultaneously performs activity recognition and user identification using smartphone sensor data. By leveraging shared representations and task-specific information, the network enhances accuracy. It combines a shared convolutional neural network (CNN) with bidirectional long short-Term memory (BiLSTM) layers to extract features from raw sensor data. The CNN identifies patterns, while the BiLSTM captures sequential dependencies and contextual information. These features are processed by fully connected layers for both activity recognition and user identification. A multi-Task loss function is proposed to optimize both tasks based on their complexity and importance. The approach is tested on the UCI-HAR dataset, containing diverse activities from multiple users. Experimental results show that the proposed CNN-BiLSTM model achieves 97.64% accuracy in activity recognition and 82.59% in user identification, surpassing current state-of-The-Art methods. This demonstrates the effectiveness of the multi-Task learning framework and the advantage of using the shared CNN-BiLSTM architecture for improving both tasks. © 2025 The Author(s).

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Groundbreaking taxonomy of metaverse characteristics

引用

Artificial Intelligence Review 2025年第8期58卷 1-50页

作者： Sadeghi-Niaraki, Abolghasem Rahimi, Fatema Azlan, Nur Alya Emanuelle Binti Song, Houbing Ali, Farman Choi, Soo-Mi Department of Computer Science and Engineering and Convergence Engineering for Intelligent Drone XR Research Center Sejong University Seoul Korea Republic of BaltimoreMD21250 United States Department of Applied AI School of Convergence College of Computing and Informatics Sungkyunkwan University Seoul03063 Korea Republic of

The Metaverse, a dynamic and immersive virtual realm, has captured the imagination of researchers and enthusiasts worldwide. This survey paper aims to introduce a groundbreaking taxonomy for the characteristics of the Metaverse offering a structured and adaptable framework that extends beyond existing categorizations by incorporating dynamic transformations. Unlike prior taxonomies, which often focus on fixed attributes, our approach emphasizes the dynamic evolution of Metaverse characteristics. Through an extensive review of published literature, this study explores key technological, social, economic, and ethical dimensions of the Metaverse. It introduces a process-oriented classification based on 23 distinct characteristics, including immersification, spatiotemporalification, interactification, persistentification, presentification, personification, unification, imaginification, economification, uncertaintification, and credification. By mapping these evolving aspects, we provide a structured and future-proof foundation for understanding the Metaverse’s continuous development. This survey establishes a new standard for comprehensiveness and innovation, shedding light on the diverse facets that have been explored in literature. Through this novel taxonomy, we provide a detailed map of the current landscape and offer insights that pave the way for future research and development in this burgeoning digital frontier. © The Author(s) 2025.

关键词： Comprehensive analysis Futuristic exploration Metaverse characteristics Paradigm-shifting taxonomy Revolutionary categorization

来源：评论

学校读者我要写书评

暂无评论

LowResIR-Net: Lightweight Residual Network for ADL Recognition Using Low-Resolution Infrared Sensor Data 16

LowResIR-Net: Lightweight Residual Network for ADL Recogniti...

引用

Detecting activities of daily living (ADL) is crucial for supported living and medical monitoring. Traditional approaches rely on high-resolution sensors and computationally intensive algorithms, limiting their scalability and practical use in real-world settings. This research introduces LowResIR-Net, a compact deep residual network architecture specifically tailored for recognizing ADL utilizing low-resolution infrared sensor data. LowResIR-Net efficiently extracts key features from low-dimensional data by leveraging the advantages of infrared sensing, such as privacy protection and light resilience. Its architecture combines residual connections with depth-wise convolutions, enhancing feature learning while maintaining a compact model size. The model’s performance is evaluated using the Coventry-2018 dataset, containing low-resolution infrared data from three residential sensor locations. Our tests have shown that in all three cases, LowResIR-Net consistently achieves superior performance compared to other cutting-edge deep learning models, such as CNN, LSTM, BiLSTM, GRU, and BiGRU. LowResIR-Net attains the most excellent accuracies of 90.18%, 93.75%, and 93.30%, and F1-scores of 81.81%, 89.55%, and 88.92% for scenarios I, II, and III, respectively. Furthermore, it achieves these results with fewer parameters than most other models. LowResIR-Net’s strong performance highlights its ability to accurately capture complex patterns in low-resolution infrared data for ADL detection. This study demonstrates the potential of lightweight deep learning and low-resolution infrared sensors in developing privacy-focused and computationally efficient wellness monitoring systems. © 2025 The Author(s).

关键词： Infrared detectors

来源：评论

学校读者我要写书评

暂无评论

Exploring the fixed point theory and numerical modeling of fish harvesting system with Allee effect

引用

Modeling Earth Systems and Environment 2025年第4期11卷 1-18页

作者： Yasin, Muhammad Waqas Akhtar, Mobeen Ahmed, Nauman Akgül, Ali Al-Mdallal, Qasem Department of mathematics University of Narowal Narowal Pakistan Department of Mathematics and Statistics The University of Lahore Lahore Pakistan Department of Electronics and Communication Engineering Saveetha School of Engineering SIMATS Tamilnadu Chennai India Department of Mathematics Art and Science Faculty Siirt University Siirt 56100 Turkey Department of Mathematics Mathematics Research Center Near East University Near East Boulevard Nicosia 99138 Turkey Department of Computer Engineering Biruni University Istanbul Topkapi 34010 Turkey Department of Mathematical Sciences UAE University P.O. Box 17551 Al Ain United Arab Emirates

Fish harvesting has a major role in nutritive food that is easily accessible for human nourishment. In this article, a reaction-diffusion fish harvesting model with the Allee effect is analyzed. The study of population models is a need of this hour because by using precautionary measures, mankind can handle the issue of food better. The basic mathematical properties are studied such as equilibrium analysis, stability, and consistency of this model. The Implicit finite difference and backward Euler methods are used for the computational results of the underlying model. The linear analysis of both schemes is derived and schemes are unconditionally stable. By using the Taylor series consistency of both schemes is proved. The positivity of the Implicit finite difference scheme is proved by using the induction technique. A test problem has been used for the numerical results. For the various values of the parameters, the simulations are drawn. The dynamical properties of continuous models, like positivity, are absent from the simulations produced by the backward Euler scheme. Implicit finite difference scheme preserves the dynamical properties of the model such as positivity, consistency, and stability. Simulations of the test problem prove the effectiveness of the Implicit finite difference scheme. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025.

关键词： Allee effect Analysis Backward Euler scheme Holling type II Implicit finite difference scheme Reaction-diffusion model Simulations

来源：评论

学校读者我要写书评

暂无评论

Enhancing malware detection performance: leveraging K-Nearest Neighbors with Firefly Optimization Algorithm

引用

Multimedia Tools and Applications 2025年第12期84卷 10071-10094页

作者： Al Saaidah, Adeeb Abualhaj, Mosleh M. Shambour, Qusai Y. Abu-Shareha, Ahmad Adel Abualigah, Laith Al-Khatib, Sumaya N. Alraba’nah, Yousef H Department of Networks and Cybersecurity Al-Ahliyya Amman University Amman Jordan Department of Software Engineering Al-Ahliyya Amman University Amman Jordan Department of Data Science and Artificial Intelligence Al-Ahliyya Amman University Amman Jordan Hourani Center for Applied Scientific Research Al-Ahliyya Amman University Amman19328 Jordan Research Center University of Tabuk Tabuk71491 Saudi Arabia MEU Research Unit Middle East University Amman11831 Jordan Department of Electrical and Computer Engineering Lebanese American University Byblos13-5053 Lebanon College of Engineering Yuan Ze University Taoyuan32003 Taiwan Applied Science Research Center Applied Science Private University Amman11931 Jordan School of Engineering and Technology Sunway University Malaysia Petaling Jaya27500 Malaysia

Malware detection plays a crucial role in ensuring robust cybersecurity amidst the ever-evolving cyber threats. This research paper delves into the realm of machine learning (ML) algorithms for malware detection, with a specific emphasis on the K-Nearest Neighbors (KNN) algorithm, utilizing tailored parameter settings and the Firefly Optimization Algorithm (FOA). The study leverages the MalMem-2022 dataset to assess the efficacy of KNN and KNN with FOA in malware detection. The impact of parameter tuning and feature selection on classification is elucidated by comparing the performance of both approaches. Encouragingly, the results reveal promising advancements in one of the multiclass classification scenarios when employing KNN with FOA, yielding noteworthy enhancements in Accuracy, Recall, Precision, Matthews Correlation Coefficient, TNR, and F1-score by 2.65%, 2.65%, 2.47%, 3.59%, 0.17%, and 2.61%, respectively. These findings accentuate the significance of optimizing KNN’s parameters and implementing FOA for feature selection, culminating in heightened accuracy in malware detection. The symbiotic fusion of KNN and FOA exhibits remarkable effectiveness in augmenting the classification model’s performance, thus offering valuable insights for bolstering cybersecurity measures. Adopting ML algorithms, particularly KNN with specific parameter settings and FOA, presents a promising avenue for enhancing malware detection capabilities. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.

关键词： Nearest neighbor search

来源：评论

学校读者我要写书评

暂无评论

Cultivating Cluster Enactment in Wireless Sensor Networks Through Hybrid Metaheuristics Intended for Energy Optimization

引用

SN computer Science 2025年第3期6卷 1-12页

作者： Kushwaha, Ajay Goswami, Anjali Sharada, K. Sharma, Manmohan Berar, Gagandeep Edara, Deepak Chowdary Department of Computer Science and Engineering Rungta College of Engineering and Technology R1 CG Bhilai India Department of Mathematics and Statistics College of Science and Theoretical Studies Saudi Electronic University Riyadh 13323 Saudi Arabia Department of Computer Science and Engineering GITAM (Deemed to be University) AP Visakhapatnam India Department of Computer Science and Engineering Manipal University Jaipur Rajasthan Jaipur India Department of Computer Science and Engineering Chitkara University Institute of Engineering and Technology Chitkara University Punjab Rajpura India Department of CSE Vignan’s Foundation for Science Technology and Research Vadlamudi AP Guntur India

Though traditional search methods shine when faced with local goals and a plethora of decision factors, they could struggle when faced with engineering design limitations. On the other hand, metaheuristic approaches promise to find global optima in real-world optimization problems, even in nonlinear and multimodal settings. In order to improve cluster head selection in Wireless Sensor Networks (WSNs), this research presents a new metaheuristic method that combines the principles of Cuckoo Search method (CSA) and Selection by Optimization (SBO). When choosing a Chief Cluster Head (CCH), the CSA algorithm takes hop count into account, whereas SBO finds the best cluster heads by considering communication range. By comparing the suggested approach to both stand-alone SBO and CSA methods, we can see how well it performs in terms of throughput, latency, energy usage, delivery ratio, and packet overhead. The combined SBO-CSA method improves cluster performance, simplifies cluster head selection in WSNs, and increases network efficiency overall, according to the results. To prove its efficacy in enhancing network performance, the suggested solution outperforms standalone SBO and CSA approaches across a range of performance parameters, such as throughput, latency, energy consumption, and packet overhead. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2025.

关键词： CCH CH CSA Metaheuristic methods SBO

来源：评论

学校读者我要写书评

暂无评论

Correction to: Innovative numerical method for solving heat conduction using subdivision collocation (Journal of applied mathematics and Computing, (2025), 10.1007/s12190-025-02429-9)

引用

Journal of applied mathematics and Computing 2025年 1-2页

作者： Malik, Safia Tehmina Ejaz, Syeda Rezapour, Shahram Inc, Mustafa Mustafa, Ghulam Department of Mathematics The Government Sadiq College Women University Bahawalpur Bahawalpur Pakistan Department of Medical Research China Medical University Hospital China Medical University Taichung Taiwan Department of Mathematics Firat University Elazig23119 Turkey Department of Mathematics Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Saveetha University Tamil Nadu Chennai602105 India Department of Computer Engineering Biruni University Ghulam Mustafa Istanbul34010 Turkey Department of Mathematics The Islamia University of Bahawalpur Bahawalpur63100 Pakistan

Typesetter introduced error in author Ghulam Mustafa affiliation and needs to be correctly read as: department of mathematics, The Islamia University of Bahawalpur, 63,100, Pakistan. Original article has been correcte...

关键词：

来源：评论

学校读者我要写书评

暂无评论

Energy-Efficient Cloud Computing Through Reinforcement Learning-Based Workload Scheduling

引用

International Journal of Advanced computer Science and Applications 2025年第4期16卷 645-656页

作者： Malipatil, Ashwini R Paramasivam, M.E. Gulyamova, Dilfuza Saravanan, Aanandha Ramesh, Janjhyam Venkata Naga Muniyandy, Elangovan Ghodhbani, Refka Department of Computer Science and Engineering BNM Institute of Technology Bangalore India Department of ECE Sona College of Technology Salem India Computer Engineering Department University of Information Technologies in Tashkent Tashkent Uzbekistan Department of ECE Vel Tech Rangarajan Dr.Sagunthala R&D Institute of Science and Technology Chennai India Department of CSE Graphic Era Hill University Dehradun248002 India Department of CSE Graphic Era Deemed to be University Uttarakhand Dehradun248002 India Department of Biosciences-Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Chennai India Applied Science Research Center Applied Science Private University Amman Jordan Center for Scientific Research and Entrepreneurship Northern Border University Arar73213 Saudi Arabia

The basis for current digital infrastructure is cloud computing, which allows for scalable, on-demand computational resource access. Data center power consumption, however, has skyrocketed because of demand increases, raising operating costs and their footprint. Traditional workload scheduling algorithms often assign performance and cost priority over energy efficiency. This paper proposes a workload scheduling method utilizing deep reinforcement learning (DRL) that adjusts dynamically according to present cloud situations to ensure optimal energy efficiency without compromising performance. The proposed method utilizes Deep Q-Networks (DQN) to perform feature engineering to identify key workload parameters such as execution time, CPU and memory consumption, and subsequently schedules tasks smartly based on these results. Based on evaluation output, the model brings down the latency to 15 ms and throughput up to 500 tasks/sec with 92% efficiency in load balancing, 95% resource usage, and 97% QoS. The proposed approach yields improved performance in terms of key parameters compared to conventional approaches such as Round Robin, FCFS, and heuristic methods. These findings show how reinforcement learning can significantly enhance the scalability, reliability, and sustainability of cloud environments. Future work will focus on enhancing fault tolerance, incorporating federated learning for decentralized optimization, and testing the model on real-world multi-cloud infrastructures. © (2025), (Science and Information Organization). All Rights Reserved.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Temperature Effects on Microalgae Cell Growth in an Outdoor HLTP 2nd

Temperature Effects on Microalgae Cell Growth in an Outdo...

引用

2nd International Conference on Mathematical Analysis and Applications in Modeling, ICMAAM 2023

作者： Nahar, Nurun Monir, Mohammad Iftekhar Islam, Md. Rashedul Deb, Ujjwal Kumar Department of Mathematics Chittagong University of Engineering and Technology Chattogram Bangladesh Department of Mathematics Premier University Chattogram Bangladesh Jamal Nazrul Islam Research Center for Mathematical and Physical Sciences University of Chittagong Chittagong Bangladesh Department of Computer Science and Engineering International Islamic University Chittagong Chattogram Bangladesh

ISBN: (纸本)9789819791934

Biofuel from microalgae is a promising alternative that can substitute fossil fuels because they are renewable and eco-friendly. The growth of microalgae is influenced by numerous culture parameters, and temperature is considered one of the vital growth factors among them. In this study, we developed a computational growth model related to irradiance related temperature for an outdoor operated Horizontal Loop Tubular Photobioreactor (HLTP) for a specific geographic location. The effects of direct and diffuse solar radiation on the Photobioreactor (PBR) temperature are considered in this model. The meteorological data associated with the geographic location has been collected for the simulation. We simulated the model by COMSOL Multiphysics and ob-served a temperature fluctuation between 24.85 ∘C and 38.45 ∘C throughout the PBR domain. The significant issue of temperature control is also suggested from this study. Over the simulated period, a uniform cell growth has been observed. The hydrodynamic behaviors of the suspension flow inside the reactor, radial and axial temperature distribution are also analyzed in this study. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Microalgae

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：