Article
Computer Science, Artificial Intelligence
Daniel Gibert, Jordi Planes, Carles Mateu, Quan Le
Summary: This paper presents a hybrid approach that combines deep learning and hand-crafted features for malware classification, achieving state-of-the-art performance in experiments.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Shaojie Yang, Yongjun Wang, Haoran Xu, Fangliang Xu, Mantun Chen
Summary: This study proposed a framework based on contrastive learning to reduce the impact of past knowledge and pretrain the model without the participation of labels. The method achieved high accuracy in malware identification and multiclass detection, outperforming supervised models in limited labeled samples.
COMPUTERS & SECURITY
(2022)
Article
Computer Science, Hardware & Architecture
Yongchao Zhang, Zhe Liu, Yu Jiang
Summary: Researchers have achieved great success in the automatic classification and detection of malware using machine learning methods in recent years. However, most approaches rely heavily on training samples, making it difficult to identify new malware families not included in the training set. In this study, a soft relevance value (s-value) is proposed as a new way to evaluate feature soft relevance using the mixed distance criterion. Experimental results demonstrate that the proposed method achieves a balance in accuracy, training, and prediction time, outperforming state-of-the-art machine learning approaches and capable of identifying new malware families.
IEEE TRANSACTIONS ON RELIABILITY
(2022)
Article
Engineering, Multidisciplinary
Ahmet Cagdas Seckin, Mine Seckin
Summary: A new feature extraction method for fabric defect detection is proposed, which is faster and more accurate compared to traditional texture feature extraction methods. This method can be used on low-level devices.
ALEXANDRIA ENGINEERING JOURNAL
(2022)
Article
Computer Science, Artificial Intelligence
Jieren Cheng, Jiachen Zheng, Xiaomei Yu
Summary: This paper provides an overview of the prevalent methods for detecting malicious codes, including signature-based, behavioral-based, and machine learning approaches. The effective malicious features are summarized and novel machine learning methods are discussed in depth. Furthermore, an ensemble interpretable framework is explored for automatic and efficient detection of malicious codes.
INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Tal Tsafrir, Aviad Cohen, Etay Nir, Nir Nissim
Summary: We are currently in an era where video files are widely used for communication and advertising, but many users are not aware of the risks associated with malicious video files. Cyber-criminals have taken advantage of this and used MP4 files to launch cyberattacks due to its vulnerabilities. Antivirus software solutions are limited in detecting unknown malware, so machine learning algorithms have been effective in detecting known and unknown malware in various formats, domains, and platforms. In this paper, three innovative and efficient feature extraction methodologies for unknown MP4 file malware detection are presented and evaluated against a collection of 6,229 files. The best performing configuration outperforms state-of-the-art feature extraction methodologies in terms of detection and generalization capabilities.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Mohammed M. Alani, Ali Ismail Awad
Summary: The Android operating system is widely used, but faces security challenges due to the rapid adoption. This paper presents a lightweight Android malware detection system based on explainable machine learning, achieving an accuracy exceeding 98%.
Article
Computer Science, Information Systems
Oguz Emre Kural, Erdal Kilic, Ceyda Aksac
Summary: Due to Android's popularity, it has become a target for cybercriminals who constantly introduce malware specifically tailored for user routines. This paper proposes an audio-based malware family detection approach that converts Android applications to audio files and extracts audio-based features. Through experiments with different classifiers, it was found that effective malware family classification can be achieved with a small number of audio-based features.
Article
Computer Science, Artificial Intelligence
Tianlei Wang, Jiuwen Cao, Xiaoping Lai, Q. M. Jonathan Wu
Summary: Autoencoding is an important branch of representation learning in deep neural networks, and the newly developed WSI-AE and OCC algorithm based on it were experimentally proven to be effective in comparison with state-of-the-art AEs and OCC algorithms.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2021)
Article
Computer Science, Information Systems
Beenish Urooj, Munam Ali Shah, Carsten Maple, Muhammad Kamran Abbasi, Sidra Riasat
Summary: Android has become a favorite target for hackers due to its popularity, making it a challenge for security providers to detect and identify malware embedded in Android applications. Machine learning approaches have emerged as a more effective way to tackle the complexity and originality of Android threats. This research paper proposes a model that incorporates innovative static feature sets and uses machine learning algorithms to detect vulnerabilities in Smartphone applications, achieving a high accuracy rate and low false positive rate.
Article
Automation & Control Systems
Wei Yuan, Yuan Jiang, Heng Li, Minghui Cai
Summary: This article focuses on on-device Android malware detection, proposing a lightweight detector based on the broad learning method. The detector mainly uses one-shot computation for model training, achieving higher accuracy than shallow learning models and approaching deep learning models.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2021)
Article
Computer Science, Information Systems
Anson Pinhero, M. L. Anupama, P. Vinod, C. A. Visaggio, N. Aneesh, S. Abhijith, S. AnanthaKrishnan
Summary: With the rapid growth of malware, automatic classification faces challenges, this study explores a new approach combining malware visualization and deep learning classification, successfully improving classification accuracy and efficiency.
COMPUTERS & SECURITY
(2021)
Article
Computer Science, Information Systems
Durmus Ozkan Sahin, Sedat Akleylek, Erdal Kilic
Summary: This study presents a framework for Android malware detection based on permissions, using multiple linear regression methods. Application permissions, critical for the security of the Android operating system, are extracted through static analysis, and machine learning techniques are employed for security analysis. Two classifiers are proposed for permission-based Android malware detection, which are compared with basic machine learning techniques on different datasets. The bagging method is utilized to increase classification performance. The results show remarkable performances with classification algorithms based on linear regression models without the need for complex algorithms.
Article
Computer Science, Theory & Methods
Junyang Qiu, Jun Zhang, Wei Luo, Lei Pan, Surya Nepal, Yang Xiang
Summary: Deep Learning (DL) is a disruptive technology that has revolutionized cyber security research, especially in the detection and classification of Android malware. While offering many advantages, DL faces challenges such as choice of architecture, feature extraction, and obtaining high-quality data.
ACM COMPUTING SURVEYS
(2021)
Article
Computer Science, Hardware & Architecture
Zheng Wang, Yang Guo, Doug Montgomery
Summary: This study proposes a method for improving the detection of malware using machine learning algorithms and feature engineering techniques on domain name distance metrics. Experimental results demonstrate that the proposed method achieves a detection accuracy of over 99% for tested domain generation algorithms.
COMPUTERS & ELECTRICAL ENGINEERING
(2022)
Article
Engineering, Electrical & Electronic
Yue Wu, Xidao Hu, Yue Zhang, Maoguo Gong, Wenping Ma, Qiguang Miao
Summary: In this paper, a skip-attention based correspondence filtering network (SACF-Net) is proposed for point cloud registration. It utilizes a feature interaction mechanism and attention mechanism to extract high-quality correspondences from different resolutions of the encoder, leading to unprecedented performance improvements on indoor and outdoor scene datasets.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Computer Science, Software Engineering
Chaoneng Li, Guanwen Feng, Yiran Jia, Yunan Li, Jian Ji, Qiguang Miao
Summary: Due to the advancement of wireless sensor and location technologies, a significant amount of mobile agent trajectory data is now available. The authors propose an unsupervised reconstruction error-based trajectory anomaly detection (RETAD) method for vehicles, which addresses the limitations of conventional anomaly detection methods. RETAD utilizes an autoencoder based on recurrent neural networks to reconstruct the original vehicle trajectories. Experimental results demonstrate that RETAD outperforms traditional distance-based, density-based, and machine learning classification algorithms in detecting anomalies.
INTERNATIONAL JOURNAL OF DATA WAREHOUSING AND MINING
(2023)
Article
Computer Science, Artificial Intelligence
Hongkai Lin, Wentian Xin, Shun Chang, Qianxue Yang, Qiguang Miao, Ruyi Liu, Liang Chang
Summary: This paper proposes a novel network structure, SWHF-Net, to address the issues in semantic segmentation, including underutilization of backbone-derived features and mismatch between small objects and large-scale encodings. SWHF-Net consists of ST-FPM and HF2M modules, which utilize feature transformation and hierarchical fusion to improve the semantic representation of multi-scale objects and enhance computational efficiency.
Article
Computer Science, Artificial Intelligence
Huan Gao, Qiguang Miao, Daikai Ma, Ruyi Liu
Summary: This paper introduces a deep mutual learning strategy to address the extreme data imbalance in brain tumor segmentation. The proposed method combines transformer layers in both the encoder and decoder of a U-Net architecture. Experimental results show that the method achieves significant performance gain over existing methods.
Article
Computer Science, Artificial Intelligence
Junmei Feng, Kunwei Wang, Qiguang Miao, Yue Xi, Zhaoqiang Xia
Summary: This paper proposes a hybrid-feedback collaborative filtering model that addresses the absence problem of negative feedback in the Bayesian personalized ranking (BPR) model by jointly exploiting explicit and implicit feedback. The model successfully extracts both implicit and explicit feedback features, and achieves competitive performance on public datasets.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Engineering, Electrical & Electronic
Yue Wu, Yue Zhang, Xiaolong Fan, Maoguo Gong, Qiguang Miao, Wenping Ma
Summary: Point cloud registration is a crucial problem in computer vision for applications in robotics and autopilot. This paper proposes a learning-based approach called INENet, which utilizes a threshold prediction network and a probability estimation network to find the overlapping area of point clouds. The advantages of this approach include automatic threshold calculation, information fusion, and easy integration, as demonstrated by experimental results.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Computer Science, Software Engineering
Qianwen Chao, Pengfei Liu, Yi Han, Yingying Lin, Chaoneng Li, Qiguang Miao, Xiaogang Jin
Summary: An all-in-one traffic simulator that considers the complex behaviors of all potential road users in a realistic urban environment is urgently needed. In this work, a novel and extensible method using the force-based concept is proposed to build a heterogeneous traffic simulation. The effectiveness of this approach is demonstrated through many simulation experiments and comparisons to real-world traffic data and popular microscopic simulators.
IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Xin Feng, Guang Jia, Jiaming Peng, Liyu Huang, Xiaofeng Liang, Haoran Zhang, Yanjun Liu, Bo Zhang, Yifei Zhang, Meng Sun, Peng Li, Qiguang Miao, Ying Wang, Li Xi, Kai Hu, Tanping Li, Hui Hui, Jie Tian
Summary: This study developed a multi-exponential relaxation spectral analysis method to separately measure the Neel and Brownian relaxation times in the magnetization recovery process in pulsed vascular MPI.
Article
Computer Science, Artificial Intelligence
Wentian Xin, Ruyi Liu, Yi Liu, Yu Chen, Wenxin Yu, Qiguang Miao
Summary: Skeleton-based action recognition is a popular and important research topic in computer vision, aiming to classify human behaviors accurately through analyzing the characteristics of human joints with deep learning technology. Skeleton data has several advantages and is especially suitable for deep learning research in low-resource environments.
Article
Environmental Sciences
Xiangzeng Liu, Ziyao Wang, Jinting Wan, Juli Zhang, Yue Xi, Ruyi Liu, Qiguang Miao
Summary: In this paper, a novel model named RoadFormer is proposed to accurately detect and extract roads using remote sensing technology. The model adopts a Swin Transformer as the backbone to effectively extract long-range information, and incorporates innovative bottleneck modules and a lightweight decoder to enhance feature representation and generate extraction results. Experimental results demonstrate the advantages of RoadFormer over comparable methods on the Deepglobe and Massachusetts datasets.
Article
Environmental Sciences
Yue Xi, Wenjing Jia, Qiguang Miao, Junmei Feng, Xiangzeng Liu, Fei Li
Summary: Benefiting from the advances in object detection in remote sensing, this study proposes a Collaborative Deraining Network (CoDerainNet) that simultaneously trains a deraining subnetwork and a droneDet subnetwork to improve the accuracy of object detection in rainy weather conditions (Rainy DroneDet). Additionally, a Collaborative Teaching paradigm (ColTeaching) is introduced to remove rain-specific interference and improve detection performance. Experimental results show that CoDerainNet can reduce computational costs while maintaining comparable detection performance to state-of-the-art models.
Article
Computer Science, Artificial Intelligence
Yunan Li, Tianyu Qi, Zhuoqi Ma, Dou Quan, Qiguang Miao
Summary: Gesture recognition has received significant attention for its wide range of applications. Previous works have focused on distinguishing different gesture classes, while ignoring the impact of within-class differences caused by gesture-irrelevant factors. In multimodal gesture recognition, fusion of features or scores is a common choice, but it often leads to redundancy in gesture-relevant features across different modalities. To address these issues, a hierarchical gesture prototype framework is proposed to highlight gesture-relevant features such as poses and motions. The framework consists of a sample-level prototype and a modal-level prototype, which effectively avoids the influence of irrelevant factors and leverages the complementarity of modalities.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Environmental Sciences
Xiangzeng Liu, Xueling Xu, Xiaodong Zhang, Qiguang Miao, Lei Wang, Liang Chang, Ruyi Liu
Summary: This paper proposes a scale and rotate transform prediction net to alleviate the effect of large geometric distortion in multimodal remote sensing image registration. The image scale regression module is constructed to reduce the scale between the reference and sensed images, and the rotation estimate module is developed to predict the rotation angles. Experimental results show the superior performance of the proposed method.
Article
Computer Science, Artificial Intelligence
Yue Wu, Peiran Gong, Maoguo Gong, Hangqi Ding, Zedong Tang, Yibo Liu, Wenping Ma, Qiguang Miao
Summary: This study proposes a novel evolving registration algorithm via evolutionary multi-task optimization, which enhances escape from local optima and improves successful registration ratio. Experimental results demonstrate that the proposed method has superior performances in terms of precision and tackling local optima.
IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTATIONAL INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Yue Wu, Yue Zhang, Wenping Ma, Maoguo Gong, Xiaolong Fan, Mingyang Zhang, A. K. Qin, Qiguang Miao
Summary: Three-dimensional point cloud registration is a significant area in computer vision. To address the challenges posed by complex scenes and incomplete observations, we propose a partial-to-partial registration network (RORNet) that extracts reliable overlapping representations from partially overlapping point clouds and uses them for registration. Our experimental results show that our method outperforms other partial registration methods.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
