Article
Computer Science, Artificial Intelligence
Huijuan Zhu, Huahui Wei, Liangmin Wang, Zhicheng Xu, Victor S. Sheng
Summary: Android has become the most popular mobile operating system due to its open source nature, wide hardware compatibility, and vast application ecosystem. However, its open source nature also makes it a prime target for malware. Existing manual feature-based malware detection methods lack effectiveness and code coverage. To address this, we propose an automated extraction method that characterizes crucial parts of the Dalvik executable into RGB images. Additionally, we introduce MADRF-CNN, a novel CNN variant that incorporates multi-scale context information to capture dependencies between different parts of the image derived from the Dex file. Experimental results demonstrate that our method achieves an accuracy of 96.9%, outperforming state-of-the-art solutions.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Cuiping Shi, Haiyang Wu, Liguo Wang
Summary: This paper proposes a double branch fusion network of CNN and enhanced graph attention network (CEGAT) based on key sample selection strategy for hyperspectral image classification. By eliminating spectral redundancy, extracting and assigning attention weight to spatial and spectral correlation features, and enhancing the relationship between nodes, the network achieves better classification performance.
Article
Computer Science, Information Systems
Wooyeol Yang, Yongsu Park
Summary: This study introduces a novel method utilizing convolution neural networks to identify symmetric-key algorithms, achieving high classification accuracy with traces extracted from Intel processor trace. The research not only identifies types of encryption algorithms, but also the number of key bits and block-cipher modes.
Article
Computer Science, Artificial Intelligence
Hui-juan Zhu, Wei Gu, Liang-min Wang, Zhi-cheng Xu, Victor S. Sheng
Summary: The popularity and flexibility of the Android platform make it a prime target for malicious attackers. By extracting permissions, API calls, and hardware features, a new malware detection framework called MSerNetDroid is proposed. The framework utilizes a novel architectural unit, Multi-Head Squeeze-and-Excitation Residual block (MSer), to learn the correlation between features and recalibrate them from multiple perspectives. Experimental results show that MSerNetDroid successfully detects malware with an accuracy of 96.48%, outperforming state-of-the-art approaches.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Hardware & Architecture
Kutub Thakur, Hamed Alqahtani, Gulshan Kumar
Summary: The intelligent system IDGADS is capable of quickly detecting algorithmically generated domains with 99% accuracy based on easy-to-compute features of real domain name system (DNS) traffic. It can serve as the first line of defense in a security stack for validating DNS queries.
COMPUTERS & ELECTRICAL ENGINEERING
(2021)
Article
Computer Science, Artificial Intelligence
Puneet Kumar, Shalini Batra, Balasubramanian Raman
Summary: The study surveys traditional and meta-heuristic approaches for optimizing deep neural networks and proposes a genetic algorithm-based method for hyper-parameter and data subset optimization, achieving significant speedups in computation time.
Article
Chemistry, Multidisciplinary
Abigail Copiaco, Leena El Neel, Tasnim Nazzal, Husameldin Mukhtar, Walid Obaid
Summary: This study introduces an innovative all-in-one malware identification model that efficiently classifies malware across diverse file types by utilizing pre-trained neural network models and grayscale transform-based features, reducing computational load.
APPLIED SCIENCES-BASEL
(2023)
Article
Multidisciplinary Sciences
TianYue Liu, HongQi Zhang, HaiXia Long, Jinmei Shi, YuHua Yao
Summary: This study proposes a new method for Android malware classification, called BIR-CNN, which combines CNN, batch normalization, and inception-residual network modules, achieving high accuracy rates. Experimental results demonstrate the effectiveness of this model in classification of Android malware, especially in malware category and family classification.
SCIENTIFIC REPORTS
(2022)
Article
Computer Science, Artificial Intelligence
Zhimeng Han, Muwei Jian, Gai-Ge Wang
Summary: This paper proposes an efficient model called ConvUNeXt, based on the classical UNet, for medical image segmentation with a low number of parameters. The model incorporates improvements such as larger convolution kernels, depth-wise separable convolution, residual connections, and a lightweight attention mechanism to enhance segmentation performance. Experimental results demonstrate superior performance compared to the standard UNet, particularly with limited data.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Geochemistry & Geophysics
Kaiwen Zheng, Jie Huang, Man Zhou, Danfeng Hong, Feng Zhao
Summary: In this work, we propose an uncertainty-aware adaptive pansharpening network (UAPN) that integrates PAN information spatial variantly to restore LMS information with an uncertainty mechanism. Experimental results demonstrate the superiority of our UAPN with fewer parameters and FLOPs, outperforming other state-of-the-art methods both qualitatively and quantitatively on multiple satellite datasets.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Multidisciplinary Sciences
Xiaojuan Wang, Yuntao Wei
Summary: A CT sequence image edge segmentation optimization algorithm was proposed in this study to address the high failure rate and low accuracy in computed tomography image edge segmentation. By clustering pixels in the CT sequence image space, calculating similarity using Euclidean distance, and hierarchically optimizing convolution neural network, the algorithm achieved high recognition rate and accuracy.
Article
Computer Science, Artificial Intelligence
Limin Shen, Jiayin Feng, Zhen Chen, Zhongkui Sun, Dongkui Liang, Hui Li, Yuying Wang
Summary: To accurately identify and classify Android malware by category and family, this paper proposes a method called SelAttConvLstm. The method converts network traffic flows into grayscale images and uses a deep learning model, SelAttConvLstm, to detect malicious Android APPs. The model considers both spatial and temporal features of network flow and incorporates self-attention weights to improve performance. Experimental results demonstrate the effectiveness of the method in detecting and classifying malware.
APPLIED INTELLIGENCE
(2023)
Article
Engineering, Multidisciplinary
M. Kavitha, R. Gayathri, Kemal Polat, Adi Alhudhaif, Fayadh Alenezi
Summary: This paper introduces an enhanced CNN method for hyperspectral image classification, aiming to improve classification accuracy by merging convolutional layer outputs and using a 1 x 1 convolution layer for feature extraction.
Article
Multidisciplinary Sciences
Niharika Das, Sujoy Das
Summary: Cardiac magnetic resonance imaging (CMRI) is a non-invasive imaging technique used to analyze the structure and function of the heart, providing functional information for diagnosing and managing cardiovascular disease. CMRI image segmentation provides quantification parameters and the manual segmentation process is time-consuming and subjective. This study utilizes a convolutional neural network model for segmentation tasks, optimizing parameters and establishing the relationship between the epoch hyperparameter and accuracy, achieving an accuracy of 0.88.
Article
Optics
Yin Wang, Jiaqing Zhao
Summary: Digital image correlation (DIC) is a non-contact optical method that tracks the speckle pattern on the specimen surface to calculate displacement and strain. Traditional DIC methods have limitations, but deep learning-based DIC methods (Deep-DIC) show promising performance. In this paper, a new Hermite dataset and a new network architecture designed for DIC tasks are proposed, and their superiority over other Deep-DIC methods is demonstrated.
OPTICS AND LASERS IN ENGINEERING
(2023)
Article
Computer Science, Interdisciplinary Applications
Wei Li, Gai-Ge Wang
Summary: EHO and BBO are two intelligent algorithms that proposed the BLEHO algorithm by interconnecting and learning based on biogeography. The algorithm dynamically changes the topological structure of the elephant population, uses different update and separation operators, and preserves a certain number of individuals through an elitism strategy to ensure better evolutionary process.
ENGINEERING WITH COMPUTERS
(2022)
Article
Computer Science, Information Systems
Kai Zhang, Jiao Tian, Hongwang Xiao, Ying Zhao, Wenyu Zhao, Jinjun Chen
Summary: Blockchain has attracted attention from the IoT research community due to its decentralization and consistency. However, the accessibility of all nodes to the chain data raises privacy concerns. To address this issue, we propose a novel LDP mechanism that splits and perturbs input numerical data using digital bits, without requiring a fixed input range and large data volume. Our adaptive privacy budget allocation model significantly reduces the deviation of the perturbation function and provides high data utility while maintaining privacy.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
Haijuan Zhang, Gai-Ge Wang
Summary: This paper proposes a TC_NSGAIII algorithm that combines centroid distance and transfer learning to solve the challenging problem of finding the changing Pareto front quickly and accurately in dynamic multi-objective optimization problems. Experimental results demonstrate the effectiveness of the proposed algorithm, showing significant improvement in performance.
COMPLEX & INTELLIGENT SYSTEMS
(2023)
Article
Computer Science, Software Engineering
Zhaoyu Xue, Wanwan Guo, Zhihua Cui, Wensheng Zhang
Summary: This paper proposes a collaborative computation-offloading model to solve the problem of limited resources in user terminals in the Internet of Things. By offloading tasks and using a global evaluation strategy, the model aims to optimize model execution time, task execution time, energy consumption, and device workload.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Article
Computer Science, Software Engineering
Tao Ye, Wenting Li, Jiangjiang Zhang, Zhihua Cui
Summary: This study introduces a multi-objective software defect prediction model and an immune optimization algorithm to address the challenges of data imbalance and parameter selection in software defect prediction. By optimizing the defect detection rate and defect false alarm rate, better prediction results are achieved.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Article
Automation & Control Systems
Xingjuan Cai, Yang Lan, Zhixia Zhang, Jie Wen, Zhihua Cui, Wensheng Zhang
Summary: In this article, a skin cancer detection model based on federated learning integrated with deep generation model is proposed to address the problem of data insufficiency and data source privacy in healthcare IoT. The model utilizes dual generative adversarial networks and knee point-driven evolutionary algorithm to improve the quality of generated images and protect patient information privacy. Experimental results show high accuracy and area under the curve for the proposed model.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Computer Science, Software Engineering
Tian Fan, Wanwan Guo, Zhixia Zhang, Zhihua Cui
Summary: With the rapid development of big data, the growth of data promotes the progress of the Internet of Things (IoT). To solve the instability of MEC performance and the conflict of interest between users and service providers, the paper proposes a virtual machine migration model based on many-objective optimization. The results of simulations show the effectiveness and superiority of the proposed approach compared to other algorithms.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Article
Computer Science, Software Engineering
Zhixia Zhang, Jie Wen, Xingjuan Cai, Zhihua Cui
Summary: This paper proposes a many-objective optimization algorithm based on dual criteria and mixed distribution correction strategy. The algorithm addresses challenges faced by existing algorithms, such as domination resistance and dimensional crisis, and achieves significant advantages in maintaining the convergence and diversity of the population.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Article
Computer Science, Artificial Intelligence
Zhuoxuan Lan, Binquan Zhang, Jie Wen, Zhihua Cui, Xiao-Zhi Gao
Summary: This paper proposes a malicious code detection model based on sequential three-way decision to solve the problem that traditional two-way decision based methods fail to consider the influence of decision making under the condition of insufficient information in dynamic environments with complex and massive data. The model introduces sequential three-way decision into the domain of malicious code to avoid the risk of possible error detection due to insufficient information. Furthermore, a multi-objective sequential three-way decision model is designed to improve the overall performance of the detection model and eliminate the subjectivity of threshold selection. The simulation results show that the model guarantees detection performance and improves decision efficiency effectively, fitting better in real dynamic detection environments.
APPLIED INTELLIGENCE
(2023)
Article
Computer Science, Software Engineering
Zhigang Zhang, Zhixia Zhang, Zhihua Cui
Summary: With the increasing adoption of IoT, malware targeting vulnerable IoT devices has become a major concern. This study proposes a novel federated malware detection framework based on many-objective optimization (FMDMO) to address the challenges in malware hunting in IoT. It provides a cross-platform compatible basis with privacy protection and enhances training efficiency while maintaining cross-architectural generalization.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2023)
Article
Computer Science, Information Systems
Muneeb Ul Hassan, Mubashir Husain Rehmani, Jinjun Chen
Summary: Over the past decade, blockchain technology has gained significant attention due to its integration with various everyday applications of modern information and communication technologies (ICT). The peer-to-peer (P2P) architecture of blockchain enhances these applications by providing strong security and trust-oriented guarantees. However, recent research has shown that blockchain networks may still face security, privacy, and reliability issues. In this article, we provide a comprehensive survey on the integration of anomaly detection models in blockchain technology. We discuss the role of anomaly detection in ensuring security, present evaluation metrics and requirements, survey various models, and highlight future research directions.
IEEE COMMUNICATIONS SURVEYS AND TUTORIALS
(2023)
Article
Computer Science, Artificial Intelligence
Ying Zhao, Dong Yuan, Jia Tina Du, Jinjun Chen
Summary: Directional distribution analysis is essential for abstracting dispersion and orientation of spatial datasets, but it must be used cautiously to protect individuals' privacy. There is a tension between accurate directional distribution results and location privacy. In this paper, we propose a geo-ellipse-indistinguishability privacy notion to protect individual location data and present elliptical privacy mechanisms based on gamma distribution and multivariate normal distribution. The empirical evaluation shows that our proposed elliptical approach achieves significantly higher directional distribution utility compared to circular noise function based methods.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Cybernetics
Xingjuan Cai, Wanwan Guo, Mengkai Zhao, Zhihua Cui, Jinjun Chen
Summary: This article proposes a knowledge graph-based many-objective model for explainable social recommendation (KGMESR), which considers the explainability, accuracy, novelty, and content quality of social recommendation results. The model utilizes social behavior information to calculate user similarity and quantifies the explainability of results using entity vectors and embedding vectors. A many-objective recommendation algorithm based on the partition deletion strategy is designed for efficiency. Experimental results demonstrate preferable recommendation results and two case studies affirm the explainability of the proposed model.
IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS
(2023)
Article
Engineering, Multidisciplinary
Wei Wan, Gaige Wang, Junyu Dong
Summary: The study introduces an enhanced version of the Adaptive Cross-Generation Differential Evolution algorithm called SIACGDE, which improves algorithm performance by strengthening the initialization strategy and optimizing parameters. Experimental results show that the algorithm outperforms others in terms of diversity and convergence.
CMES-COMPUTER MODELING IN ENGINEERING & SCIENCES
(2022)