检索结果-内蒙古大学图书馆

Application of human activity/action recognition: a review

Multimedia Tools and Applications 2025年 1-30页

作者： sedaghati, Nazanin ardebili, Sondos Ghaffari, Ali Department of Computer Engineering Tabriz Branch Islamic Azad University Tabriz Iran Department of Computer Engineering Faculty of Engineering and Natural Science Istinye University Istanbul Turkey Department of Computer Science Khazar University Baku Azerbaijan

Human activity recognition is a crucial domain in computer science and artificial intelligence that involves the Detection, Classification, and Prediction of human activities using sensor data such as accelerometers, gyroscopes, etc. This field utilizes time-series signals from sensors present in smartphones and wearable devices to extract human activities. Various types of sensors, including inertial HAR sensors, physiological sensors, location sensors, cameras, and temporal sensors, are employed in diverse environments within this domain. It finds valuable applications in various areas such as smart homes, elderly care, the Internet of Things (IoT), personal care, social sciences, rehabilitation engineering, fitness, and more. With the advancement of computational power, deep learning algorithms have been recognized as effective and efficient methods for detecting and solving well-established HAR issues. In this research, a review of various deep learning algorithms is presented with a focus on distinguishing between two key aspects: activity and action. Action refers to specific, short-term movements and behaviors, while activity refers to a set of related, continuous affairs over time. The reviewed articles are categorized based on the type of algorithms and applications, specifically sensor-based and vision-based. The total number of reviewed articles in this research is 80 sources, categorized into 42 references. By offering a detailed classification of relevant articles, this comprehensive review delves into the analysis and scrutiny of the scientific community in the HAR domain using deep learning algorithms. It serves as a valuable guide for researchers and enthusiasts to gain a better understanding of the advancements and challenges within this field. © The Author(s) 2025.

关键词： Human Action Detection Human Activity Recognition (HAR) Human–computer Interaction (HCI) Sensors Smartphones

来源：评论

学校读者我要写书评

暂无评论

Local Search-Based Anytime Algorithms for Continuous Distributed Constraint Optimization Problems

引用

IEEE/CAA Journal of Automatica Sinica 2025年第1期12卷 288-290页

作者： Xin Liao Khoi Hoang Xin Luo the College of Computer and Information Science Southwest University the Department of Computer Science and Engineering Washington University in St.Louis IEEE

Dear Editor,The distributed constraint optimization problems(DCOPs) [1]-[3]provide an efficient model for solving the cooperative problems of multi-agent systems, which has been successfully applied to model the real-world problems like the distributed scheduling [4], sensor network management [5], [6], multi-robot coordination [7], and smart grid [8]. However, DCOPs were not well suited to solve the problems with continuous variables and constraint cost in functional form, such as the target tracking sensor orientation [9], the air and ground cooperative surveillance [10], and the sensor network coverage [11].

关键词：

来源：评论

学校读者我要写书评

暂无评论

Predicting Distant Metastatic Sites in Cancer Using miRNA and mRNA Expression Data

引用

Current Bioinformatics 2025年第5期20卷 473-485页

作者： Mamatkarimov, Dostonjon Kang, Jiahui Han, Kyungsook Department of Electrical and Computer Engineering Inha University Incheon Korea Republic of

Background: Cancer patients with metastasis face a much lower survival rate and a higher risk of recurrence than those without metastasis. So far, several learning methods have been proposed to predict cancer metastasis, but most of these methods are intended to predict lymph node metastasis rather than distant metastasis. Distant metastasis is more difficult to predict than lymph node metastasis because distant metastasis is detected after a comprehensive examination of the entire body, and there are not enough publicly available tumor samples with distant metastasis that can be used for training learning methods. Predicting distant metastatic sites is even more challenging than predicting whether distant metastasis will occur or not. Methods: The problem of predicting distant metastatic sites is a multi‐class and multi‐label classification problem;there are more than two classes for distant metastatic sites (bone, brain, liver, lung, and other organs), and a single sample can have multiple labels for multiple metastatic sites. We transformed the multi‐label and multi‐class problem into multiple single‐label binary problems. For each metastatic site, we built a random forest model that deals with binary classification and linked the models along a chain. Results: Testing the model on miRNA and mRNA expression datasets of several cancer types showed a high performance in all performance measures. In the comparison of our model with other methods, our method outperformed the others. Conclusion: We developed a new method for predicting multiple metastatic sites using miRNA and mRNA expression data. The technique will be useful in predicting distant metastatic sites before distant metastasis occurs, which in turn will help clinicians determine treatment options for cancer patients. © 2025 Bentham science Publishers.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

Knowledge distillation on spatial-temporal graph convolutional network for traffic prediction

引用

International Journal of computers and Applications 2025年第1期47卷 45-56页

作者： Izadi, Mohammad Safayani, Mehran Mirzaei, Abdolreza Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan Iran

Efficient real-time traffic prediction is crucial for reducing transportation time. To predict traffic conditions, we employ a spatio-temporal graph neural network (ST-GNN) to model our real-time traffic data as temporal graphs. Despite its capabilities, it often encounters challenges in delivering efficient real-time predictions for real-world traffic data. Recognizing the significance of timely prediction due to the dynamic nature of real-time data, we employ knowledge distillation (KD) as a solution to enhance the execution time of ST-GNNs for traffic prediction. In this paper, We introduce a cost function designed to train a network with fewer parameters (the student) using distilled data from a complex network (the teacher) while maintaining its accuracy close to that of the teacher. We use knowledge distillation, incorporating spatial-temporal correlations from the teacher network to enable the student to learn the complex patterns perceived by the teacher. However, a challenge arises in determining the student network architecture rather than considering it inadvertently. To address this challenge, we propose an algorithm that utilizes the cost function to calculate pruning scores, addressing small network architecture search issues, and jointly fine-tunes the network resulting from each pruning stage using KD. Ultimately, we evaluate our proposed ideas on two real-world datasets, PeMSD7 and PeMSD8. The results indicate that our method can maintain the student's accuracy close to that of the teacher, even with the retention of only (Formula presented.) of network parameters. © 2024 Informa UK Limited, trading as Taylor & Francis Group.

关键词： Cost functions

来源：评论

学校读者我要写书评

暂无评论

A Two-Level Neural-RL-Based Approach for Hierarchical Multiplayer Systems Under Mismatched Uncertainties

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2025年第3期6卷 759-772页

作者： Zhong, Xiangnan Ni, Zhen Florida Atlantic University Department of Electrical Engineering and Computer Science Boca RatonFL33431 United States

AI and reinforcement learning (RL) have attracted great attention in the study of multiplayer systems over the past decade. Despite the advances, most of the studies are focused on synchronized decision-making to attain Nash equilibrium, where all the players take actions simultaneously. On the other hand, however, in complex applications, certain players may have an advantage in making sequential decisions and this situation introduces a hierarchical structure and influences how other players respond. The control design for such system is challenging since it relies on solving the coupled Hamilton-Jacobi equation. The situation becomes more difficult when the learning process is exposed to complex uncertainties with unreliable data being exchanged. Therefore, in this article, we develop a new learning-based control approach for a class of nonlinear hierarchical multiplayer systems subject to mismatched uncertainties. Specifically, we first formulate this new problem as a multiplayer Stackelberg-Nash game in conjunction with the hierarchical robust-optimal transformation. Theoretical analysis confirms the equivalence of this transformation and ensures that the designed control policies can achieve stable equilibrium. Then, a two-level neural-RL-based approach is developed to automatically and adaptively learn the solutions. The stability of this online learning process is also provided. Finally, two numerical examples are presented to demonstrate the effectiveness of the developed learning-based robust control design. © 2024 IEEE.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Informative task classification with concatenated embeddings using deep learning on crisisMMD

引用

International Journal of computers and Applications 2025年第2期47卷 123-140页

作者： Jain, Tarun Gopalani, Dinesh Kumar Meena, Yogesh Department of Computer Science and Engineering Manipal University Jaipur Jaipur India Department of Computer Science and Engineering Malaviya National Institute of Technology Jaipur India

Disastrous situations pose a formidable challenge, testing our resilience against nature's fury and the race against time to prevent the loss of human life. It is noted that in such situations that Microblogging platforms like Twitter(now X) have proven to be an invaluable resource for citizens to share the real-time updates in their locality. Building upon this premise, we are proposing a new hybrid robust approach that integrates transfer learning with selective features from Liwc and manually engineered features to achieve state-of-the-art results on the CrisisMMD dataset on the informative task. In our innovative approach, we harnessed the power of Deep learning combined with traditional ML practices to improve the model, we utilized OpenAI embeddings and minute details from the tweets like the sentiment, text polarity, Retweet address, links included etc. We employed a dense-dropout-based concatenate model for further processing. We established baselines of each technique individually as well as commonly used techniques like Tf-Idf, BoW, Glove embeddings. Notably, our proposed model exhibits a significant accuracy improvement, achieving an accuracy of 87.93%, surpassing the highest accuracy of traditional methods of 82.93% obtained with Tf-Idf and ensemble voting. Through this work, we conclude that there is a scope of improvement in disaster-related NLP work with the introduction of transformer-based architecture and Pretrained GPT which include a massive corpus of text improving the semantic coherency which in turn lead to higher evaluation metrics. This work in future can be extended to incorporate different embeddings as well as using techniques like Multi-headed attention as well as RNNs like GST, GRU, etc. © 2025 Informa UK Limited, trading as Taylor & Francis Group.

关键词： Tweets

来源：评论

学校读者我要写书评

暂无评论

Benchmarking deep learning for wetland mapping in Denmark using remote sensing data

引用

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 2025年 18卷 11953-11962页

作者： Asif, Muhammad Rizwan Department of Electrical and Computer Engineering Aarhus University Aarhus N Denmark

Wetlands are critical ecosystems providing numerous ecological services, yet they face significant threats from human activities and climate change. Therefore, accurate mapping and monitoring of wetlands are crucial for formulating effective conservation and restoration strategies. While remote sensing combined with deep learning offers a promising solution, inconsistencies in wetland classification systems - where different regions define wetland types based on their policy frameworks and conservation priorities - limit the applicability of these models. Such inconsistencies make it difficult to assess their limitations in different contexts. Notably, no study has yet leveraged deep learning for mapping wetlands within Denmark's unique wetland classification system, as defined by the Danish Nature Conservation framework. Therefore, this study presents a comprehensive benchmark analysis of several deep learning models for wetland classification in Denmark. We utilize publicly available high-resolution multispectral aerial imagery and digital elevation models and evaluate the performance of three well-established network architectures: Fully Convolutional Network, U-Net, and DeepLabV3. We also assess the impact of incorporating near-infrared and digital elevation model data in addition to traditional optical imagery. The results show that DeepLabV3 model outperforms other models, particularly when additional data layers are included, achieving the highest overall accuracy and F-measure score. Our findings also reveal that while deep learning models can effectively classify certain wetlands, challenges remain in distinguishing wetland with ecological similarities and in handling noisy labels. This benchmark provides a foundation for future work aimed at improving deep learning methods for wetland mapping in Denmark. © 2025 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Simulation Tools for Cognitive Radio Network: A Survey

引用

IEEE Access 2025年 13卷 1-27页

作者： Islam, A. K. M. Muzahidul Bin Rabbani, Rakin Aadeeb, Md. Shadman Zareei, Mahdi Mansoor, Nafees Zubaer, Molla Md. United International University Department of Computer Science and Engineering Dhaka1212 Bangladesh BRAC University Department of Computer Science and Engineering Dhaka1212 Bangladesh Tecnologico de Monterrey School of Engineering and Sciences Monterrey64849 Mexico University of Liberal Arts Bangladesh Department of Computer Science and Engineering Dhaka1209 Bangladesh Department of Electrical Electronic and Communication Engineering Dhaka1216 Bangladesh

Due to recent expansion of wireless communications, it has become impossible to cope with the allotment of the precious spectrum while resources for wireless communication are bounded and finite. Hence, the cognitive radio concept is introduced to ensure optimal utilization of these invaluable resources. By emerging research in the field of cognitive radio network, fresh researchers are looking for a suitable simulator for the cognitive radio network since the implementation and testbed is not practical and feasible invariably. Moreover, there is no simulator that has been designed with cognitive radio networks in mind. Furthermore, this research area is relatively new. Therefore, often it becomes very hard to decide which network simulator to work with, as the availability of reference is significantly low. As a result, much effort is required to make a proper decision. In this paper, we take a detailed look at current simulators used by researchers in the field of cognitive radio networks to help researchers identify appropriate tools for simulation. To further aid researchers, we classify the simulation tools, demonstrate their architectures, describe features, elucidate advantages and disadvantages of these tools, and report some open issues and challenges for future research. © 2025 The Authors.

关键词： Cognitive radio

来源：评论

学校读者我要写书评

暂无评论

Medicare fraud detection: a comparative study on the effectiveness of one-class and binary classification models

引用

International Journal of computers and Applications 2025年第3期47卷 323-328页

作者： Leevy, Joffrey L. Hancock, John Khoshgoftaar, Taghi M. Department of Electrical Engineering and Computer Science Florida Atlantic University Boca RatonFL United States

Research into Medicare fraud detection that utilizes machine learning methodologies is of great national interest due to the significant fiscal ramifications of this type of fraud. Our big data analysis pertains to three Medicare datasets: Part B, Part D, and DMEPOS. In this paper, One-Class Classification (OCC) and binary classification approaches are evaluated with six different classifiers. We note that when a practitioner is faced with a high cost or unreasonable wait associated with acquiring a second label for binary classification, OCC may intuitively seem like the better alternative. The evaluative metrics used in our study are Area Under the Receiver Operating Characteristic Curve (AUC) and Area Under the Precision-Recall Curve (AUPRC). As a result of our investigation, binary classification was found to be more effective than OCC in detecting Medicare fraud. Additionally, we determined that gradient-boosting algorithms, specifically CatBoost and XGBoost, yield the most robust performance. © 2024 Informa UK Limited, trading as Taylor & Francis Group.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Noisy Computing of the Threshold Function 36

Noisy Computing of the Threshold Function

引用

36th International Conference on Algorithmic Learning Theory, ALT 2025

作者： Wang, Ziao Ghaddar, Nadim Zhu, Banghua Wang, Lele Department of Electrical Engineering and Computer Science University of Michigan Ann Arbor United States Department of Electrical and Computer Engineering University of Toronto Canada Department of Electrical and Computer Engineering University of Washington United States Department of Electrical and Computer Engineering University of British Columbia Canada

Coping with noise in computing is an important problem to consider in large systems. With applications in fault tolerance (Hastad et al., 1987;Pease et al., 1980;Pippenger et al., 1991), noisy sorting (Shah and Wainwright, 2018;Agarwal et al., 2017;Falahatgar et al., 2017;Heckel et al., 2019;Wang et al., 2024a;Gu and Xu, 2023;Kunisky et al., 2024), noisy searching (Berlekamp, 1964;Horstein, 1963;Burnashev and Zigangirov, 1974;Pelc, 1989;Karp and Kleinberg, 2007), among many others, the goal is to devise algorithms with the minimum number of queries that are robust enough to detect and correct the errors that can happen during the computation. In this work, we consider the noisy computing of the threshold-k function. For n Boolean variables x = (x1, ..., xn) ∈ {0, 1}n, the threshold-k function THk(·) computes whether the number of 1's in x is at least k or not, i.e., (Equation presented) The noisy queries correspond to noisy readings of the bits, where at each time step, the agent queries one of the bits, and with probability p, the wrong value of the bit is returned. It is assumed that the constant p ∈ (0, 1/2) is known to the agent. Our goal is to characterize the optimal query complexity for computing the THk function with error probability at most δ. This model for noisy computation of the THk function has been studied by Feige et al. (1994), where the order of the optimal query complexity is established;however, the exact tight characterization of the optimal number of queries is still open. In this paper, our main contribution is tightening this gap by providing new upper and lower bounds for the computation of the THk function, which simultaneously improve the existing upper and lower bounds. The main result of this paper can be stated as follows: for any 1 ≤ k ≤ n, there exists an algorithm that computes the THk function with an error probability at most δ = o(1), and the algorithm uses at most (Equation presented) queries in expectation. Here we define m (Eq

关键词： Active learning

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：