检索结果-内蒙古大学图书馆

Differential Image-Based Scalable YOLOv7-Tiny Implementation for Clustered Embedded Systems

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2024年第11期25卷 16036-16047页

作者： Hong, Sunghoon Park, Daejin Kyungpook Natl Univ Dept Elect & Elect Engn Daegu 41566 South Korea Kyungpook Natl Univ Sch Elect Engn Daegu 41566 South Korea

Convolutional neural networks (CNNs) for powerful visual image analysis are gaining popularity in artificial intelligence. The main difference in CNNs compared to other artificial neural networks is that many convolutional layers are added, which improve the performance of visual image analysis by extracting the feature maps required for image classification. However, algorithm optimization is required to run applications that require low-latency in edge compute modules with limited processing resources. In this paper, we propose a novel algorithm optimization method for fast CNNs by using continuous differential images. The main idea is to reduce computation variably by using the differential value of the input in each convolutional layer. Also, the proposed method is compatible with all types of CNNs, and the performance is better when the pixel value difference of continuous images is low. We use the DarkNet framework to evaluate our algorithm using fast convolution and half convolution approaches on a clustered system. As a result, when the input frame rate is 10 fps, FLOPs are reduced by about 4.92 times compared to the original YOLOv7-tiny. By reducing the FLOPs of the convolutional layer, the inference speed increases to about 4.86 FPS, performing 1.57 times faster than the original YOLOv7-tiny. In the case of parallel processing that used two edge compute modules for using half convolution approach, FLOPs reduced more, and the response speed improved. In addition, faster Object detection implementation is possible by additionally expanding up to 7 compute modules in a scalable clustered embedded system as much as the user wants.

关键词： Detectors Convolution Accuracy Computational complexity Real-time systems Feature extraction Convolutional neural networks classification algorithms Graphics processing units Embedded systems clustered systems deep learning fast convolution

来源：评论

学校读者我要写书评

暂无评论

CBCTL-IDS: A Transfer Learning-Based Intrusion Detection System Optimized With the Black Kite Algorithm for IoT-Enabled Smart Agriculture

引用

IEEE ACCESS 2025年 13卷 46601-46615页

作者： Zhou, Hai Zou, Haojie Zhou, Pinxi Shen, Yue Li, Di Li, Wei Hunan Agr Univ Coll Informat & Intelligence Changsha 410128 Peoples R China

Smart Agriculture(SA) stands as a critical application frontier for the evolution of the IoT, catalyzing the transformation of traditional agricultural practices into modern, data-driven, intelligent, and automated systems. This transition not only elevates the levels of automation and intelligence in agricultural production but also introduces a spectrum of cybersecurity challenges. In particular, the widespread deployment of interconnected devices within these systems poses significant risks to the security and integrity of data. As a pivotal defense mechanism against malicious activities and anomalous traffic, Network Intrusion Detection Systems (NIDS) play an indispensable role in ensuring the security of cyberspace. Nevertheless, existing methods for detecting anomalous traffic in IoT networks still exhibit limitations in accuracy, thereby hindering the efficiency of identifying potential threats. To enhance the capability of threat detection, this paper proposes a novel network intrusion detection method, CBCTL-IDS, based on transfer learning. This method integrates three core components: Convolutional Neural Networks (CNN), the Black Kite Algorithm (BKA), and a Confidence Averaging mechanism. By incorporating pre-trained models such as MobileNet, EfficientNet, Xception, VGG19, and Inception into the domain of IoT intrusion detection, CBCTL-IDS leverages transfer learning to achieve efficient feature extraction and classification, addressing the limitations of traditional methods in feature representation. Furthermore, the BKA is utilized to adaptively optimize the hyperparameters of multiple DL models, significantly enhancing their classification performance. Through ensemble learning, the top three performing models are selected, and the Confidence Averaging mechanism is applied to further improve the stability and reliability of detection results. Experimental results demonstrate that the CBCTL-IDS method achieves detection accuracy rates exceeding 99% on t

关键词： Security Smart agriculture Accuracy Convolutional neural networks Transfer learning Computer crime Wireless sensor networks Production classification algorithms Image edge detection Intrusion detection system smart agriculture Internet of Things transfer learning

来源：评论

学校读者我要写书评

暂无评论

A Weighted classification Method Based on Adaptive Feature Selection

引用

IEEE ACCESS 2022年 10卷 58635-58646页

作者： Ni, Ruizheng Qiu, Ruichang Luo, Zhiwei Chen, Jie Jin, Zheming Liu, Zhigang Beijing Jiaotong Univ Sch Elect Engn Beijing 100044 Peoples R China Beijing Inst Space Launch Technol Beijing 100076 Peoples R China

This paper proposes a new classification method, which can adaptively select effective features and give different classification weights according to the classification ability of features, this method effectively eliminates the influence of invalid features in classification process. For the classification problem of fault diagnosis, this method judges the fault type from the azimuth relative to the normal state in multi-dimensional space, and judges the fault degree from the ratio of distance, the method can not only give the fault category, but also give the fault probability, fault degree and whether there is a new fault type. In the process of updating the classification center, the classification method proposed in this paper solves the problem that the location of the classification center changes with the sample processing order. The effectiveness of this method is tested by the transformer vibration data (TVIB) and five data sets published in MATLAB and UCI, and compared with K-Mean, self-organizing map (SOM) methods, it is proved that considering the classification weight can effectively improve the classification accuracy, and the proposed method has strong applicability for different classification problems and good processing ability for large capacity and multi feature data.

关键词： classification algorithms Transformers Mathematical models Support vector machines Reactive power Fault diagnosis Vibrations classification method classification features fault diagnosis transformer vibration

来源：评论

学校读者我要写书评

暂无评论

Ultrafast Photodiodes With High Saturation Power Based on an Automated Intelligent Design System

引用

IEEE JOURNAL OF QUANTUM ELECTRONICS 2025年第1期61卷

作者： Huang, Junjing Duan, Xiaofeng Liu, Kai Huang, Yongqing Ren, Xiaomin Beijing Univ Posts & Telecommun State Key Lab Informat Photon & Opt Commun Beijing 100876 Peoples R China

The rapid development of inverse design algorithms and the increasing availability of high-speed computing resources have made it possible to exploit heuristic algorithms to assist the design of photonic devices. This paper presents an automated intelligent design system for semiconductor photodetectors, which is capable of identifying the optimal performance of the photodetector under arbitrary material distribution. The aforementioned method was employed for the design of a modified uni-traveling carrier photodetector (MUTC-PD), exhibiting a bandwidth of 232GHz(@-5V) and an RF output power of 16.799dBm(@100GHz). In comparison to the original structure and traditional manual scanning methods, the bandwidth of this device has been increased by 96GHz and 25GHz, respectively, which serves to illustrate the effectiveness and superiority of this approach. Furthermore, the system is also instrumental in the discovery of new material combinations and structural designs, thereby driving innovation and advancement in the field of semiconductor photodetector technology.

关键词： Bandwidth Heuristic algorithms Doping Particle swarm optimization Optimization Iterative methods Photodiodes classification algorithms Capacitance Wireless communication Automated intelligent design system photodectors velocity-varying climbing particle swarm optimization

来源：评论

学校读者我要写书评

暂无评论

Face Recognition Method for Underground Engineering Based on Dual-Target Domain Adaptation and Discriminative Feature Learning

引用

IEEE ACCESS 2025年 13卷 24223-24236页

作者： Yu, Yongqiang Guo, Cong Fan, Lidan Zhang, Jiyun Yu, Liwei Li, Peitao Henan Polytech Univ Sch Civil Engn Jiaozuo 454003 Peoples R China Henan Polytech Univ Henan Key Lab Underground Engn Disaster Prevent Jiaozuo 454003 Peoples R China Henan Railway Construct & Investment Grp Co Ltd Zhengzhou 450000 Peoples R China

The poor lighting and dust in underground engineering environments often induce serious noise and motion blur to face recognition cameras, decreasing the recognition accuracy even under Near-InfraRed (NIR) mode. Although various methods to improve algorithm robustness can enhance the overall model performance, the face recognition accuracy remains insufficient for the large number of blurred and NIR images generated during underground engineering. Accordingly, this paper proposes a novel face recognition method for underground engineering environments based on dual-target domain adaptation (DTDA) and discriminative feature learning (DFL). The targeting of the algorithm to blurred and NIR images is enhanced through DTDA, while the inter-class feature variability and intra-class feature compactness for domain adaptation face recognition are ensured by DFL. Extensive experiments are conducted on public and self-built datasets to verify the proposed method. The results show that (1) dual domain adversarial loss, dual CORAL loss, Blur-to-Sharp and NIR-to-VIS image reconstruction, and face keypoints intermediate domain can improve the DTDA performance of the model. (2) Dedicated encoders, center loss, and triplet loss can improve the DFL ability of the model. (3) The high Rank-1 accuracies and area under curve (AUC) values of the Receiver Operating Characteristic (ROC) curves on the three domains indicate that the proposed face recognition method can adapt to the complex working conditions in underground engineering environments.

关键词： Face recognition Feature extraction Image reconstruction Adaptation models Representation learning Accuracy Robustness classification algorithms Information filters Personnel Dual-target domain adaptation discriminative feature learning face recognition underground engineering

来源：评论

学校读者我要写书评

暂无评论

An Object-Based Change Detection Method Considering Temporal-Spatial Similarity in Long Time Series

引用

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 2025年 63卷

作者： Chen, Lisu Li, Shanhong Zhu, Enyan Shanghai Maritime Univ Coll Ocean Sci & Engn Shanghai 201306 Peoples R China Ningbo Univ Zhejiang Collaborat Innovat Ctr Land & Marine Spat Ningbo 315211 Peoples R China Shanghai Maritime Univ Coll Transport & Commun Shanghai 201306 Peoples R China

A new object-based change detection method has been proposed to address the limitations of existing research based on pixel-based change detection, as well as the neglect of concurrent changes, pixel spatiotemporal information, and the alteration of boundary information due to changes. Additionally considering the concurrence of changes, the measurement of similarity in time series of adjacent pixels needs to take into account the phase shifting and scaling of the time series. In this study, based on the considerations mentioned above, we first constructed time series using available Landsat 5, 7, and 8 data collected in the study area and used a pixel-based change detection method to obtain change information. Then, considering the heterogeneity within objects, we first combined the change information to extract change seeds. After that, the inside_similarity metric was introduced, which is computed by dynamic time wraping (DTW) algorithm, and it was used to impose sequential constraints on the expansion of seeds. Considering that changes can alter both the interior and boundaries of objects, we applied a conditional judgment to all pixels outside the seeds. Through quantitative assessment in three experimental areas, the method proposed in this article improved producer's accuracy (PA) by 6.9%, 2.7%, and 5.5% and user's accuracy (UA) by 6.1%, 3.1%, and 6.6% with F1-score improved by 6.49%, 2.91%, and 7.45% compared to purely pixel-based change detection methods. Combined with qualitative assessment, the object-based change detection method is proved to increase the accuracy of change detection.

关键词： Time series analysis Accuracy Landsat Data mining Feature extraction classification algorithms Change detection algorithms Remote sensing Image segmentation Urban areas Change detection dynamic time wraping (DTW) flood fill object-based

来源：评论

学校读者我要写书评

暂无评论

A Nonparametric Data-Driven Classifier Based on the Cumulant Generating Function

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 2025年 73卷 519-533页

作者： Kay, Steven Adhikari, Kaushallya Tang, Bo Univ Rhode Isl Dept Elect Comp & Biomed Engn Kingston RI 02881 USA Worcester Polytech Inst Dept Elect & Comp Engn Worcester MA 01609 USA

We introduce a nonparametric data-driven classifier that harnesses the statistical properties of the data through the cumulant generating function of the training data. Its implementation is straightforward, requiring only a single tuning parameter. Moreover, it ensures global solutions due to inherent convex optimization. The classifier is explainable, where unexpected or poor results can be interpreted and ameliorated. We derive the properties of the classification statistic, offering insightful observations. We apply the classifier to real-world datasets. The simulation results demonstrate the efficacy of the proposed classifier in signal classification, even in scenarios with mismatched training and testing datasets. Moreover, the results demonstrate that the CGFC has lower computational complexity compared to neural networks.

关键词： Training data Probability density function Maximum likelihood estimation Vectors Smoothing methods Training classification algorithms Testing Laplace equations Tuning Cumulant generating function data-driven classifier neural networks nonparametric classifier sonar data analysis

来源：评论

学校读者我要写书评

暂无评论

LEPRE: An Updatable Database-Dependent Range Encoding Algorithm

引用

IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS 2025年第2期43卷 551-562页

作者： Lin, Hsin-Tsung Chen, Wei-Cheng Wang, Pi-Chung Natl Chung Hsing Univ Dept Comp Sci & Engn Taichung 402 Taiwan

Packet classification is a key mechanism that classifies incoming packets into flows to enable software-defined networking as well as a variety of networking services. Currently, ternary content addressable memory (TCAM) has been widely used for high-speed and low-latency packet classification. However, both range expansion and update performance are the fundamental issues for TCAM-based packet classification. A rule containing ranges could be replicated to multiple rules after converting its ranges into prefixes or ternary strings to occupy more than one TCAM entry. Many range encoding algorithms have been proposed to alleviate or avoid the problem of range expansion. These algorithms can be classified into database-independent (DI) and database-dependent (DD). While database-independent algorithms can accommodate new ranges without re-encoding the existing ranges, they may still cause rule replication. In contrast, database-dependent algorithms could avoid rule replication by adaptively encoding ranges, but new ranges may result in updates of the existing ranges. Accordingly, both types of algorithms may multiply the cost of TCAM updates. In this paper, we propose a DD range-encoding algorithm, Longest Enclosure Prefix Range Encoding (LEPRE), which can ensure that any new range does not cause any rule replication and re-encoding of the existing ranges. LEPRE employs the original fields as a part of range encoding to significantly decrease the requirements of extra bits for range encoding. Our experiment results show that LEPRE can maximize the TCAM storage efficiency. LEPRE also fully supports incremental updates to minimize the latency of TCAM updates.

关键词： Encoding Costs classification algorithms Reflective binary codes Codes Routing IP networks Associative memory Vectors Urban areas Software-defined networking (SDN) ternary CAMs range encoding

来源：评论

学校读者我要写书评

暂无评论

Android Malware Detection Based on Informative Syscall Subsequences

引用

IEEE ACCESS 2024年 12卷 189180-189190页

作者： Surendran, Roopak Uddin, Md. Meraj Thomas, Tony Pradeep, Gokul Digital Univ Kerala Kerala Secur Audit & Assurance Ctr Thiruvananthapuram 695317 Kerala India Indian Inst Informat Technol & Management Kerala Res Ctr CUSAT Thiruvananthapuram 695581 Kerala India

The Android operating system commands a dominant market share of over 70% in the smartphone industry. However, this widespread usage has resulted in a concerning increase in malware applications. While existing static malware detection mechanisms are vulnerable to code obfuscation attacks, manipulating the runtime system call (syscall) sequence remains a significant challenge for attackers. Consequently, syscall-based malware detection mechanisms are gaining prominence. Current syscall-based malware detection approaches rely on machine learning algorithms, utilizing numerical features such as syscall frequencies and transition probability matrices. However, the wide range of values in these features necessitates large datasets for effective classifier training, and susceptibility to noise and outliers persists. As a result, there is an urgent need for a binary representation of dynamic features to improve malware detection efficiency. To address this challenge, our paper proposes an innovative syscall subsequence-based binary feature representation method for machine learning-driven malware detection. By employing the information gain method, we identify informative syscall subsequences. The proposed mechanism achieves an impressive 99% accuracy in detecting malware applications using just 50% of the training data, across both the Drebin/AMD and CICMalDroid2020 datasets.

关键词： Malware Feature extraction Smart phones Training data classification algorithms Androids Machine learning algorithms Operating systems Runtime Numerical analysis Android malware system calls machine learning

来源：评论

学校读者我要写书评

暂无评论

Efficient and Generalized Image-Based CNN Algorithm for Multi-Class Malware Detection

引用

IEEE ACCESS 2024年 12卷 104317-104332页

作者： Liu, Yajun Fan, Hong Zhao, Jianguang Zhang, Jianfang Yin, Xinxin Hebei Univ Architecture Informat Engn Coll Zhangjiakou 075000 Hebei Peoples R China

With the popularity of electronic devices, the number of malware has increased dramatically, posing a serious threat to the digital world. Accurately identifying malware has become a research focus. However, there are many difficulties in the research, such as insufficient algorithm generalization ability, unbalanced datasets, and long processing and identification times. To address these problems, this study proposes a malware detection framework (VBDN) based on a convolutional neural network (CNN). The framework incorporates data visualization, balanced adoption, data augmentation, and convolutional neural network techniques to achieve over 90% accuracy in classifying malware on all four open-source datasets. The experimental work has two other contributions: first, it not only focuses on the overall recognition effect of the algorithm during the research process, but also on the recognition effect of each category with the help of a confusion matrix, which provides useful information for cybersecurity personnel, researchers, and others to carry out subsequent targeted research. Secondly, the balanced approach adopted in this paper has the following advantages: no need to construct a new dataset, consumes fewer hardware resources, automatically evaluates the sampling weights, etc. Additionally, to enhance the generalization ability of the algorithm and alleviate the overfitting problem, this paper employs data augmentation techniques to improve the adopted method. By comparing with several state-of-the-art algorithms, it can be observed that the VBDN framework proposed in this paper achieves the desired results in time with acceptable accuracy.

关键词： Malware Feature extraction Biological neural networks Convolutional neural networks classification algorithms Deep learning Mathematical models Data visualization Malware detection convolutional neural network (CNN) data visualization balanced adoption data enhancement

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：