Current spatio-temporal action detection methods lack sufficient capabilities in extracting and comprehending spatio-temporal information. This paper introduces an end-to-end Adaptive Cross-Scale Fusion Encoder-Decode...Current spatio-temporal action detection methods lack sufficient capabilities in extracting and comprehending spatio-temporal information. This paper introduces an end-to-end Adaptive Cross-Scale Fusion Encoder-Decoder (ACSF-ED) network to predict the action and locate the object efficiently. In the Adaptive Cross-Scale Fusion Spatio-Temporal Encoder (ACSF ST-Encoder), the Asymptotic Cross-scale Feature-fusion Module (ACCFM) is designed to address the issue of information degradation caused by the propagation of high-level semantic information, thereby extracting high-quality multi-scale features to provide superior features for subsequent spatio-temporal information modeling. Within the Shared-Head Decoder structure, a shared classification and regression detection head is constructed. A multi-constraint loss function composed of one-to-one, one-to-many, and contrastive denoising losses is designed to address the problem of insufficient constraint force in predicting results with traditional methods. This loss function enhances the accuracy of model classification predictions and improves the proximity of regression position predictions to ground truth objects. The proposed method model is evaluated on the popular dataset UCF101-24 and JHMDB-21. Experimental results demonstrate that the proposed method achieves an accuracy of 81.52% on the Frame-mAP metric, surpassing current existing methods.展开更多
As Deepfake technology continues to evolve,the distinction between real and fake content becomes increasingly blurred.Most existing Deepfake video detectionmethods rely on single-frame facial image features,which limi...As Deepfake technology continues to evolve,the distinction between real and fake content becomes increasingly blurred.Most existing Deepfake video detectionmethods rely on single-frame facial image features,which limits their ability to capture temporal differences between frames.Current methods also exhibit limited generalization capabilities,struggling to detect content generated by unknown forgery algorithms.Moreover,the diversity and complexity of forgery techniques introduced by Artificial Intelligence Generated Content(AIGC)present significant challenges for traditional detection frameworks,whichmust balance high detection accuracy with robust performance.To address these challenges,we propose a novel Deepfake detection framework that combines a two-stream convolutional network with a Vision Transformer(ViT)module to enhance spatio-temporal feature representation.The ViT model extracts spatial features from the forged video,while the 3D convolutional network captures temporal features.The 3D convolution enables cross-frame feature extraction,allowing the model to detect subtle facial changes between frames.The confidence scores from both the ViT and 3D convolution submodels are fused at the decision layer,enabling themodel to effectively handle unknown forgery techniques.Focusing on Deepfake videos and GAN-generated images,the proposed approach is evaluated on two widely used public face forgery datasets.Compared to existing state-of-theartmethods,it achieves higher detection accuracy and better generalization performance,offering a robust solution for deepfake detection in real-world scenarios.展开更多
Health monitoring of electro-mechanical actuator(EMA)is critical to ensure the security of airplanes.It is difficult or even impossible to collect enough labeled failure or degradation data from actual EMA.The autoenc...Health monitoring of electro-mechanical actuator(EMA)is critical to ensure the security of airplanes.It is difficult or even impossible to collect enough labeled failure or degradation data from actual EMA.The autoencoder based on reconstruction loss is a popular model that can carry out anomaly detection with only consideration of normal training data,while it fails to capture spatio-temporal information from multivariate time series signals of multiple monitoring sensors.To mine the spatio-temporal information from multivariate time series signals,this paper proposes an attention graph stacked autoencoder for EMA anomaly detection.Firstly,attention graph con-volution is introduced into autoencoder to convolve temporal information from neighbor features to current features based on different weight attentions.Secondly,stacked autoencoder is applied to mine spatial information from those new aggregated temporal features.Finally,based on the bench-mark reconstruction loss of normal training data,different health thresholds calculated by several statistic indicators can carry out anomaly detection for new testing data.In comparison with tra-ditional stacked autoencoder,the proposed model could obtain higher fault detection rate and lower false alarm rate in EMA anomaly detection experiment.展开更多
In order to enhance the accuracy of Air Traffic Control(ATC)cybersecurity attack detection,in this paper,a new clustering detection method is designed for air traffic control network security attacks.The feature set f...In order to enhance the accuracy of Air Traffic Control(ATC)cybersecurity attack detection,in this paper,a new clustering detection method is designed for air traffic control network security attacks.The feature set for ATC cybersecurity attacks is constructed by setting the feature states,adding recursive features,and determining the feature criticality.The expected information gain and entropy of the feature data are computed to determine the information gain of the feature data and reduce the interference of similar feature data.An autoencoder is introduced into the AI(artificial intelligence)algorithm to encode and decode the characteristics of ATC network security attack behavior to reduce the dimensionality of the ATC network security attack behavior data.Based on the above processing,an unsupervised learning algorithm for clustering detection of ATC network security attacks is designed.First,determine the distance between the clustering clusters of ATC network security attack behavior characteristics,calculate the clustering threshold,and construct the initial clustering center.Then,the new average value of all feature objects in each cluster is recalculated as the new cluster center.Second,it traverses all objects in a cluster of ATC network security attack behavior feature data.Finally,the cluster detection of ATC network security attack behavior is completed by the computation of objective functions.The experiment took three groups of experimental attack behavior data sets as the test object,and took the detection rate,false detection rate and recall rate as the test indicators,and selected three similar methods for comparative test.The experimental results show that the detection rate of this method is about 98%,the false positive rate is below 1%,and the recall rate is above 97%.Research shows that this method can improve the detection performance of security attacks in air traffic control network.展开更多
With the rapid urbanization and exponential population growth in China,two-wheeled vehicles have become a popular mode of transportation,particularly for short-distance travel.However,due to a lack of safety awareness...With the rapid urbanization and exponential population growth in China,two-wheeled vehicles have become a popular mode of transportation,particularly for short-distance travel.However,due to a lack of safety awareness,traffic violations by two-wheeled vehicle riders have become a widespread concern,contributing to urban traffic risks.Currently,significant human and material resources are being allocated to monitor and intercept non-compliant riders to ensure safe driving behavior.To enhance the safety,efficiency,and cost-effectiveness of traffic monitoring,automated detection systems based on image processing algorithms can be employed to identify traffic violations from eye-level video footage.In this study,we propose a robust detection algorithm specifically designed for two-wheeled vehicles,which serves as a fundamental step toward intelligent traffic monitoring.Our approach integrates a novel convolutional and attention mechanism to improve detection accuracy and efficiency.Additionally,we introduce a semi-supervised training strategy that leverages a large number of unlabeled images to enhance the model’s learning capability by extracting valuable background information.This method enables the model to generalize effectively to diverse urban environments and varying lighting conditions.We evaluate our proposed algorithm on a custom-built dataset,and experimental results demonstrate its superior performance,achieving an average precision(AP)of 95%and a recall(R)of 90.6%.Furthermore,the model maintains a computational efficiency of only 25.7 GFLOPs while achieving a high processing speed of 249 FPS,making it highly suitable for deployment on edge devices.Compared to existing detection methods,our approach significantly enhances the accuracy and robustness of two-wheeled vehicle identification while ensuring real-time performance.展开更多
The basic theory of YOLO series object detection algorithms is discussed, the dangerous driving behavior dataset is collected and produced, and then the YOLOv7 network is introduced in detail, the deep separable convo...The basic theory of YOLO series object detection algorithms is discussed, the dangerous driving behavior dataset is collected and produced, and then the YOLOv7 network is introduced in detail, the deep separable convolution and CA attention mechanism are introduced, the YOLOv7 bounding box loss function and clustering algorithm are optimized, and the DB-YOLOv7 network structure is constructed. In the first stage of the experiment, the PASCAL VOC public dataset was utilized for pre-training. A comparative analysis was conducted to assess the recognition accuracy and inference time before and after the proposed improvements. The experimental results demonstrated an increase of 1.4% in the average recognition accuracy, alongside a reduction in the inference time by 4 ms. Subsequently, a model for the recognition of dangerous driving behaviors was trained using a specialized dangerous driving behavior dataset. A series of experiments were performed to evaluate the efficacy of the DB-YOLOv7 algorithm in this context. The findings indicate a significant enhancement in detection performance, with a 4% improvement in accuracy compared to the baseline network. Furthermore, the model’s inference time was reduced by 20%, from 25 ms to 20 ms. These results substantiate the effectiveness of the DB-YOLOv7 recognition algorithm for detecting dangerous driving behaviors, providing comprehensive validation of its practical applicability.展开更多
Social robot accounts controlled by artificial intelligence or humans are active in social networks,bringing negative impacts to network security and social life.Existing social robot detection methods based on graph ...Social robot accounts controlled by artificial intelligence or humans are active in social networks,bringing negative impacts to network security and social life.Existing social robot detection methods based on graph neural networks suffer from the problem of many social network nodes and complex relationships,which makes it difficult to accurately describe the difference between the topological relations of nodes,resulting in low detection accuracy of social robots.This paper proposes a social robot detection method with the use of an improved neural network.First,social relationship subgraphs are constructed by leveraging the user’s social network to disentangle intricate social relationships effectively.Then,a linear modulated graph attention residual network model is devised to extract the node and network topology features of the social relation subgraph,thereby generating comprehensive social relation subgraph features,and the feature-wise linear modulation module of the model can better learn the differences between the nodes.Next,user text content and behavioral gene sequences are extracted to construct social behavioral features combined with the social relationship subgraph features.Finally,social robots can be more accurately identified by combining user behavioral and relationship features.By carrying out experimental studies based on the publicly available datasets TwiBot-20 and Cresci-15,the suggested method’s detection accuracies can achieve 86.73%and 97.86%,respectively.Compared with the existing mainstream approaches,the accuracy of the proposed method is 2.2%and 1.35%higher on the two datasets.The results show that the method proposed in this paper can effectively detect social robots and maintain a healthy ecological environment of social networks.展开更多
Focusing on the sensitive behaviors of malware, such as privacy stealing and money costing, this paper proposes a new method to monitor software behaviors and detect malicious applications on Android platform. Accordi...Focusing on the sensitive behaviors of malware, such as privacy stealing and money costing, this paper proposes a new method to monitor software behaviors and detect malicious applications on Android platform. According to the theory and implementation of Android Binder interprocess communication mechanism, a prototype system that integrates behavior monitoring and intercepting, malware detection, and identification is built in this work. There are 50 different kinds of samples used in the experiment of malware detection, including 40 normal samples and 10 malicious samples. The theoretical analysis and experimental result demonstrate that this system is effective in malware detection and interception, with a true positive rate equal to 100% and a false positive rate less than 3%.展开更多
Automatic detection of student engagement levels from videos,which is a spatio-temporal classification problem is crucial for enhancing the quality of online education.This paper addresses this challenge by proposing ...Automatic detection of student engagement levels from videos,which is a spatio-temporal classification problem is crucial for enhancing the quality of online education.This paper addresses this challenge by proposing four novel hybrid end-to-end deep learning models designed for the automatic detection of student engagement levels in e-learning videos.The evaluation of these models utilizes the DAiSEE dataset,a public repository capturing student affective states in e-learning scenarios.The initial model integrates EfficientNetV2-L with Gated Recurrent Unit(GRU)and attains an accuracy of 61.45%.Subsequently,the second model combines EfficientNetV2-L with bidirectional GRU(Bi-GRU),yielding an accuracy of 61.56%.The third and fourth models leverage a fusion of EfficientNetV2-L with Long Short-Term Memory(LSTM)and bidirectional LSTM(Bi-LSTM),achieving accuracies of 62.11%and 61.67%,respectively.Our findings demonstrate the viability of these models in effectively discerning student engagement levels,with the EfficientNetV2-L+LSTM model emerging as the most proficient,reaching an accuracy of 62.11%.This study underscores the potential of hybrid spatio-temporal networks in automating the detection of student engagement,thereby contributing to advancements in online education quality.展开更多
Aiming at the difficulty of unknown Trojan detection in the APT flooding situation, an improved detecting method has been proposed. The basic idea of this method originates from advanced persistent threat （APT） atta...Aiming at the difficulty of unknown Trojan detection in the APT flooding situation, an improved detecting method has been proposed. The basic idea of this method originates from advanced persistent threat （APT） attack intents： besides dealing with damaging or destroying facilities, the more essential purpose of APT attacks is to gather confidential data from target hosts by planting Trojans. Inspired by this idea and some in-depth analyses on recently happened APT attacks, five typical communication characteristics are adopted to describe application’s network behavior, with which a fine-grained classifier based on Decision Tree and Na ve Bayes is modeled. Finally, with the training of supervised machine learning approaches, the classification detection method is implemented. Compared with general methods, this method is capable of enhancing the detection and awareness capability of unknown Trojans with less resource consumption.展开更多
A new approach for abnormal behavior detection was proposed using causality analysis and sparse reconstruction. To effectively represent multiple-object behavior, low level visual features and causality features were ...A new approach for abnormal behavior detection was proposed using causality analysis and sparse reconstruction. To effectively represent multiple-object behavior, low level visual features and causality features were adopted. The low level visual features, which included trajectory shape descriptor, speeded up robust features and histograms of optical flow, were used to describe properties of individual behavior, and causality features obtained by causality analysis were introduced to depict the interaction information among a set of objects. In order to cope with feature noisy and uncertainty, a method for multiple-object anomaly detection was presented via a sparse reconstruction. The abnormality of the testing sample was decided by the sparse reconstruction cost from an atomically learned dictionary. Experiment results show the effectiveness of the proposed method in comparison with other state-of-the-art methods on the public databases for abnormal behavior detection.展开更多
In this study,we propose a low-cost system that can detect the space outlier utilization of residents in an indoor environment.We focus on the users’app usage to analyze unusual behavior,especially in indoor spaces.T...In this study,we propose a low-cost system that can detect the space outlier utilization of residents in an indoor environment.We focus on the users’app usage to analyze unusual behavior,especially in indoor spaces.This is reflected in the behavioral analysis in that the frequency of using smartphones in personal spaces has recently increased.Our system facilitates autonomous data collection from mobile app logs and Google app servers and generates a high-dimensional dataset that can detect outlier behaviors.The density-based spatial clustering of applications with noise(DBSCAN)algorithm was applied for effective singular movement analysis.To analyze high-level mobile phone usage,the t-distributed stochastic neighbor embedding(t-SNE)algorithm was employed.These two clustering algorithms can effectively detect outlier behaviors in terms of movement and app usage in indoor spaces.The experimental results showed that our system enables effective spatial behavioral analysis at a low cost when applied to logs collected in actual living spaces.Moreover,large volumes of data required for outlier detection can be easily acquired.The system can automatically detect the unusual behavior of a user in an indoor space.In particular,this study aims to reflect the recent trend of the increasing use of smartphones in indoor spaces to the behavioral analysis.展开更多
Detecting malware on mobile devices using the Android operating system has become a critical challenge in the field of cybersecurity,in the context of the rapid increase in the number of malware variants and the frequ...Detecting malware on mobile devices using the Android operating system has become a critical challenge in the field of cybersecurity,in the context of the rapid increase in the number of malware variants and the frequency of attacks targeting Android devices.In this paper,we propose a novel intelligent computational method to enhance the effectiveness of Android malware detection models.The proposed method combines two main techniques:(1)constructing a malware behavior profile and(2)extracting features from the malware behavior profile using graph neural networks.Specifically,to effectively construct an Android malware behavior profile,this paper proposes an information enrichment technique for the function call graph of malware files,based on new graph-structured features and semantic features of the malware’s source code.Additionally,to extract significant features from the constructed behavior profile,the study proposes using the GraphSAGE graph neural network.With this novel intelligent computational method,a variety of significant features of the malware have been effectively represented,synthesized,and extracted.The approach to detecting Android malware proposed in this paper is a new study and has not been explored in previous research.The experimental results on a dataset of 40,819 Android software indicate that the proposed method performs well across all metrics,with particularly impressive accuracy and recall scores of 99.03%and 99.19%,respectively,which outperforms existing state-of-the-art methods.展开更多
To improve the detection accuracy and robustness of crowd anomaly detection,especially crowd emergency evacuation detection,the abnormal crowd behavior detection method is proposed.This method is based on the improved...To improve the detection accuracy and robustness of crowd anomaly detection,especially crowd emergency evacuation detection,the abnormal crowd behavior detection method is proposed.This method is based on the improved statistical global optical flow entropy which can better describe the degree of chaos of crowd.First,the optical flow field is extracted from the video sequences and a 2D optical flow histogram is gained.Then,the improved optical flow entropy,combining information theory with statistical physics is calculated from 2D optical flow histograms.Finally,the anomaly can be detected according to the abnormality judgment formula.The experimental results show that the detection accuracy achieved over 95%in three public video datasets,which indicates that the proposed algorithm outperforms other state-of-the-art algorithms.展开更多
Abnormal behavior detection is challenging and one of the growing research areas in computer vision.The main aim of this research work is to focus on panic and escape behavior detections that occur during unexpected/u...Abnormal behavior detection is challenging and one of the growing research areas in computer vision.The main aim of this research work is to focus on panic and escape behavior detections that occur during unexpected/uncertain events.In this work,Pyramidal Lucas Kanade algorithm is optimized using EME-HOs to achieve the objective.First stage,OPLKT-EMEHOs algorithm is used to generate the opticalflow from MIIs.Second stage,the MIIs opticalflow is applied as input to 3 layer CNN for detect the abnormal crowd behavior.University of Minnesota(UMN)dataset is used to evaluate the proposed system.The experi-mental result shows that the proposed method provides better classification accu-racy by comparing with the existing methods.Proposed method provides 95.78%of precision,90.67%of recall,93.09%of f-measure and accuracy with 91.67%.展开更多
With the continuous expansion of digital infrastructures,malicious behaviors in host systems have become increasingly sophisticated,often spanning multiple processes and employing obfuscation techniques to evade detec...With the continuous expansion of digital infrastructures,malicious behaviors in host systems have become increasingly sophisticated,often spanning multiple processes and employing obfuscation techniques to evade detection.Audit logs,such as Sysmon,offer valuable insights;however,existing approaches typically flatten event sequences or rely on generic graph models,thereby discarding the natural parent-child process hierarchy that is critical for analyzing multiprocess attacks.This paper proposes a structure-aware threat detection framework that transforms audit logs into a unified two-dimensional(2D)spatio-temporal representation,where process hierarchy is modeled as the spatial axis and event chronology as the temporal axis.In addition,entropy-based features are incorporated to robustly capture obfuscated and non-linguistic strings,overcoming the limitations of semantic embeddings.The model’s performance was evaluated on publicly available datasets,achieving competitive results with an accuracy exceeding 95%and an F1-score of at least 0.94.The proposed approach provides a promising and reproducible solution for detecting attacks with unknown indicators of compromise(IoCs)by analyzing the relationships and behaviors of processes recorded in large-scale audit logs.展开更多
With technology constantly becoming present in people’s lives, smart homes are increasing in popularity. A smart home system controls lighting, temperature, security camera systems, and appliances. These devices and ...With technology constantly becoming present in people’s lives, smart homes are increasing in popularity. A smart home system controls lighting, temperature, security camera systems, and appliances. These devices and sensors are connected to the internet, and these devices can easily become the target of attacks. To mitigate the risk of using smart home devices, the security and privacy thereof must be artificially smart so they can adapt based on user behavior and environments. The security and privacy systems must accurately analyze all actions and predict future actions to protect the smart home system. We propose a Hybrid Intrusion Detection (HID) system using machine learning algorithms, including random forest, X gboost, decision tree, K -nearest neighbors, and misuse detection technique.展开更多
In recent years,unmanned air vehicles(UAVs)are widely used in many military and civilian applications.With the big amount of UAVs operation in air space,the potential security and privacy problems are arising.This can...In recent years,unmanned air vehicles(UAVs)are widely used in many military and civilian applications.With the big amount of UAVs operation in air space,the potential security and privacy problems are arising.This can lead to consequent harm for critical infrastructure in the event of these UAVs being used for criminal or terrorist purposes.Therefore,it is crucial to promptly identify the suspicious behaviors from the surrounding UAVs for some important regions.In this paper,a novel fuzzy logic based UAV behavior detection system has been presented to detect the different levels of risky behaviors of the incoming UAVs.The heading velocity and region type are two input indicators proposed for the risk indicator output in the designed fuzzy logic based system.The simulation has shown the effective and feasible of the proposed algorithm in terms of recall and precision of the detection.Especially,the suspicious behavior detection algorithm can provide a recall of 0.89 and a precision of 0.95 for the high risk scenario in the simulation.展开更多
Quantification of behaviors in macaques provides crucial support for various scientific disciplines,including pharmacology,neuroscience,and ethology.Despite recent advancements in the analysis of macaque behavior,rese...Quantification of behaviors in macaques provides crucial support for various scientific disciplines,including pharmacology,neuroscience,and ethology.Despite recent advancements in the analysis of macaque behavior,research on multi-label behavior detection in socially housed macaques,including consideration of interactions among them,remains scarce.Given the lack of relevant approaches and datasets,we developed the Behavior-Aware Relation Network(BARN)for multi-label behavior detection of socially housed macaques.Our approach models the relationship of behavioral similarity between macaques,guided by a behavior-aware module and novel behavior classifier,which is suitable for multi-label classification.We also constructed a behavior dataset of rhesus macaques using ordinary RGB cameras mounted outside their cages.The dataset included 65?913 labels for19 behaviors and 60?367 proposals,including identities and locations of the macaques.Experimental results showed that BARN significantly improved the baseline SlowFast network and outperformed existing relation networks.In conclusion,we successfully achieved multilabel behavior detection of socially housed macaques with both economic efficiency and high accuracy.展开更多
基金support for this work was supported by Key Lab of Intelligent and Green Flexographic Printing under Grant ZBKT202301.
文摘Current spatio-temporal action detection methods lack sufficient capabilities in extracting and comprehending spatio-temporal information. This paper introduces an end-to-end Adaptive Cross-Scale Fusion Encoder-Decoder (ACSF-ED) network to predict the action and locate the object efficiently. In the Adaptive Cross-Scale Fusion Spatio-Temporal Encoder (ACSF ST-Encoder), the Asymptotic Cross-scale Feature-fusion Module (ACCFM) is designed to address the issue of information degradation caused by the propagation of high-level semantic information, thereby extracting high-quality multi-scale features to provide superior features for subsequent spatio-temporal information modeling. Within the Shared-Head Decoder structure, a shared classification and regression detection head is constructed. A multi-constraint loss function composed of one-to-one, one-to-many, and contrastive denoising losses is designed to address the problem of insufficient constraint force in predicting results with traditional methods. This loss function enhances the accuracy of model classification predictions and improves the proximity of regression position predictions to ground truth objects. The proposed method model is evaluated on the popular dataset UCF101-24 and JHMDB-21. Experimental results demonstrate that the proposed method achieves an accuracy of 81.52% on the Frame-mAP metric, surpassing current existing methods.
基金supported by National Natural Science Foundation of China(Nos.62477026,62177029,61807020)Humanities and Social Sciences Research Program of the Ministry of Education of China(No.23YJAZH047)the Startup Foundation for Introducing Talent of Nanjing University of Posts and Communications under Grant NY222034.
文摘As Deepfake technology continues to evolve,the distinction between real and fake content becomes increasingly blurred.Most existing Deepfake video detectionmethods rely on single-frame facial image features,which limits their ability to capture temporal differences between frames.Current methods also exhibit limited generalization capabilities,struggling to detect content generated by unknown forgery algorithms.Moreover,the diversity and complexity of forgery techniques introduced by Artificial Intelligence Generated Content(AIGC)present significant challenges for traditional detection frameworks,whichmust balance high detection accuracy with robust performance.To address these challenges,we propose a novel Deepfake detection framework that combines a two-stream convolutional network with a Vision Transformer(ViT)module to enhance spatio-temporal feature representation.The ViT model extracts spatial features from the forged video,while the 3D convolutional network captures temporal features.The 3D convolution enables cross-frame feature extraction,allowing the model to detect subtle facial changes between frames.The confidence scores from both the ViT and 3D convolution submodels are fused at the decision layer,enabling themodel to effectively handle unknown forgery techniques.Focusing on Deepfake videos and GAN-generated images,the proposed approach is evaluated on two widely used public face forgery datasets.Compared to existing state-of-theartmethods,it achieves higher detection accuracy and better generalization performance,offering a robust solution for deepfake detection in real-world scenarios.
基金supported by the National Natural Science Foundation of China (No.52075349)the National Natural Science Foundation of China (No.62303335)+1 种基金the Postdoctoral Researcher Program of China (No.GZC20231779)the Natural Science Foundation of Sichuan Province (No.2022NSFSC1942).
文摘Health monitoring of electro-mechanical actuator(EMA)is critical to ensure the security of airplanes.It is difficult or even impossible to collect enough labeled failure or degradation data from actual EMA.The autoencoder based on reconstruction loss is a popular model that can carry out anomaly detection with only consideration of normal training data,while it fails to capture spatio-temporal information from multivariate time series signals of multiple monitoring sensors.To mine the spatio-temporal information from multivariate time series signals,this paper proposes an attention graph stacked autoencoder for EMA anomaly detection.Firstly,attention graph con-volution is introduced into autoencoder to convolve temporal information from neighbor features to current features based on different weight attentions.Secondly,stacked autoencoder is applied to mine spatial information from those new aggregated temporal features.Finally,based on the bench-mark reconstruction loss of normal training data,different health thresholds calculated by several statistic indicators can carry out anomaly detection for new testing data.In comparison with tra-ditional stacked autoencoder,the proposed model could obtain higher fault detection rate and lower false alarm rate in EMA anomaly detection experiment.
基金National Natural Science Foundation of China(U2133208,U20A20161)National Natural Science Foundation of China(No.62273244)Sichuan Science and Technology Program(No.2022YFG0180).
文摘In order to enhance the accuracy of Air Traffic Control(ATC)cybersecurity attack detection,in this paper,a new clustering detection method is designed for air traffic control network security attacks.The feature set for ATC cybersecurity attacks is constructed by setting the feature states,adding recursive features,and determining the feature criticality.The expected information gain and entropy of the feature data are computed to determine the information gain of the feature data and reduce the interference of similar feature data.An autoencoder is introduced into the AI(artificial intelligence)algorithm to encode and decode the characteristics of ATC network security attack behavior to reduce the dimensionality of the ATC network security attack behavior data.Based on the above processing,an unsupervised learning algorithm for clustering detection of ATC network security attacks is designed.First,determine the distance between the clustering clusters of ATC network security attack behavior characteristics,calculate the clustering threshold,and construct the initial clustering center.Then,the new average value of all feature objects in each cluster is recalculated as the new cluster center.Second,it traverses all objects in a cluster of ATC network security attack behavior feature data.Finally,the cluster detection of ATC network security attack behavior is completed by the computation of objective functions.The experiment took three groups of experimental attack behavior data sets as the test object,and took the detection rate,false detection rate and recall rate as the test indicators,and selected three similar methods for comparative test.The experimental results show that the detection rate of this method is about 98%,the false positive rate is below 1%,and the recall rate is above 97%.Research shows that this method can improve the detection performance of security attacks in air traffic control network.
基金supported by the Natural Science Foundation Project of Fujian Province,China(Grant No.2023J011439 and No.2019J01859).
文摘With the rapid urbanization and exponential population growth in China,two-wheeled vehicles have become a popular mode of transportation,particularly for short-distance travel.However,due to a lack of safety awareness,traffic violations by two-wheeled vehicle riders have become a widespread concern,contributing to urban traffic risks.Currently,significant human and material resources are being allocated to monitor and intercept non-compliant riders to ensure safe driving behavior.To enhance the safety,efficiency,and cost-effectiveness of traffic monitoring,automated detection systems based on image processing algorithms can be employed to identify traffic violations from eye-level video footage.In this study,we propose a robust detection algorithm specifically designed for two-wheeled vehicles,which serves as a fundamental step toward intelligent traffic monitoring.Our approach integrates a novel convolutional and attention mechanism to improve detection accuracy and efficiency.Additionally,we introduce a semi-supervised training strategy that leverages a large number of unlabeled images to enhance the model’s learning capability by extracting valuable background information.This method enables the model to generalize effectively to diverse urban environments and varying lighting conditions.We evaluate our proposed algorithm on a custom-built dataset,and experimental results demonstrate its superior performance,achieving an average precision(AP)of 95%and a recall(R)of 90.6%.Furthermore,the model maintains a computational efficiency of only 25.7 GFLOPs while achieving a high processing speed of 249 FPS,making it highly suitable for deployment on edge devices.Compared to existing detection methods,our approach significantly enhances the accuracy and robustness of two-wheeled vehicle identification while ensuring real-time performance.
文摘The basic theory of YOLO series object detection algorithms is discussed, the dangerous driving behavior dataset is collected and produced, and then the YOLOv7 network is introduced in detail, the deep separable convolution and CA attention mechanism are introduced, the YOLOv7 bounding box loss function and clustering algorithm are optimized, and the DB-YOLOv7 network structure is constructed. In the first stage of the experiment, the PASCAL VOC public dataset was utilized for pre-training. A comparative analysis was conducted to assess the recognition accuracy and inference time before and after the proposed improvements. The experimental results demonstrated an increase of 1.4% in the average recognition accuracy, alongside a reduction in the inference time by 4 ms. Subsequently, a model for the recognition of dangerous driving behaviors was trained using a specialized dangerous driving behavior dataset. A series of experiments were performed to evaluate the efficacy of the DB-YOLOv7 algorithm in this context. The findings indicate a significant enhancement in detection performance, with a 4% improvement in accuracy compared to the baseline network. Furthermore, the model’s inference time was reduced by 20%, from 25 ms to 20 ms. These results substantiate the effectiveness of the DB-YOLOv7 recognition algorithm for detecting dangerous driving behaviors, providing comprehensive validation of its practical applicability.
基金This work was supported in part by the National Natural Science Foundation of China under Grants 62273272,62303375 and 61873277in part by the Key Research and Development Program of Shaanxi Province under Grant 2023-YBGY-243+2 种基金in part by the Natural Science Foundation of Shaanxi Province under Grants 2022JQ-606 and 2020-JQ758in part by the Research Plan of Department of Education of Shaanxi Province under Grant 21JK0752in part by the Youth Innovation Team of Shaanxi Universities.
文摘Social robot accounts controlled by artificial intelligence or humans are active in social networks,bringing negative impacts to network security and social life.Existing social robot detection methods based on graph neural networks suffer from the problem of many social network nodes and complex relationships,which makes it difficult to accurately describe the difference between the topological relations of nodes,resulting in low detection accuracy of social robots.This paper proposes a social robot detection method with the use of an improved neural network.First,social relationship subgraphs are constructed by leveraging the user’s social network to disentangle intricate social relationships effectively.Then,a linear modulated graph attention residual network model is devised to extract the node and network topology features of the social relation subgraph,thereby generating comprehensive social relation subgraph features,and the feature-wise linear modulation module of the model can better learn the differences between the nodes.Next,user text content and behavioral gene sequences are extracted to construct social behavioral features combined with the social relationship subgraph features.Finally,social robots can be more accurately identified by combining user behavioral and relationship features.By carrying out experimental studies based on the publicly available datasets TwiBot-20 and Cresci-15,the suggested method’s detection accuracies can achieve 86.73%and 97.86%,respectively.Compared with the existing mainstream approaches,the accuracy of the proposed method is 2.2%and 1.35%higher on the two datasets.The results show that the method proposed in this paper can effectively detect social robots and maintain a healthy ecological environment of social networks.
基金Supported by the National Natural Science Foundation of China(61103220)the Fundamental Research Funds for the Central Universities (6082013)+1 种基金the National Natural Science Foundation of Hubei(2011CDB456)Chenguang Program(2012710367)
文摘Focusing on the sensitive behaviors of malware, such as privacy stealing and money costing, this paper proposes a new method to monitor software behaviors and detect malicious applications on Android platform. According to the theory and implementation of Android Binder interprocess communication mechanism, a prototype system that integrates behavior monitoring and intercepting, malware detection, and identification is built in this work. There are 50 different kinds of samples used in the experiment of malware detection, including 40 normal samples and 10 malicious samples. The theoretical analysis and experimental result demonstrate that this system is effective in malware detection and interception, with a true positive rate equal to 100% and a false positive rate less than 3%.
文摘Automatic detection of student engagement levels from videos,which is a spatio-temporal classification problem is crucial for enhancing the quality of online education.This paper addresses this challenge by proposing four novel hybrid end-to-end deep learning models designed for the automatic detection of student engagement levels in e-learning videos.The evaluation of these models utilizes the DAiSEE dataset,a public repository capturing student affective states in e-learning scenarios.The initial model integrates EfficientNetV2-L with Gated Recurrent Unit(GRU)and attains an accuracy of 61.45%.Subsequently,the second model combines EfficientNetV2-L with bidirectional GRU(Bi-GRU),yielding an accuracy of 61.56%.The third and fourth models leverage a fusion of EfficientNetV2-L with Long Short-Term Memory(LSTM)and bidirectional LSTM(Bi-LSTM),achieving accuracies of 62.11%and 61.67%,respectively.Our findings demonstrate the viability of these models in effectively discerning student engagement levels,with the EfficientNetV2-L+LSTM model emerging as the most proficient,reaching an accuracy of 62.11%.This study underscores the potential of hybrid spatio-temporal networks in automating the detection of student engagement,thereby contributing to advancements in online education quality.
基金Supported by the National Natural Science Foundation of China (61202387, 61103220)Major Projects of National Science and Technology of China(2010ZX03006-001-01)+3 种基金Doctoral Fund of Ministry of Education of China (2012014110002)China Postdoctoral Science Foundation (2012M510641)Hubei Province Natural Science Foundation (2011CDB456)Wuhan Chenguang Plan Project(2012710367)
文摘Aiming at the difficulty of unknown Trojan detection in the APT flooding situation, an improved detecting method has been proposed. The basic idea of this method originates from advanced persistent threat （APT） attack intents： besides dealing with damaging or destroying facilities, the more essential purpose of APT attacks is to gather confidential data from target hosts by planting Trojans. Inspired by this idea and some in-depth analyses on recently happened APT attacks, five typical communication characteristics are adopted to describe application’s network behavior, with which a fine-grained classifier based on Decision Tree and Na ve Bayes is modeled. Finally, with the training of supervised machine learning approaches, the classification detection method is implemented. Compared with general methods, this method is capable of enhancing the detection and awareness capability of unknown Trojans with less resource consumption.
基金Project(50808025) supported by the National Natural Science Foundation of ChinaProject(20090162110057) supported by the Doctoral Fund of Ministry of Education,China
文摘A new approach for abnormal behavior detection was proposed using causality analysis and sparse reconstruction. To effectively represent multiple-object behavior, low level visual features and causality features were adopted. The low level visual features, which included trajectory shape descriptor, speeded up robust features and histograms of optical flow, were used to describe properties of individual behavior, and causality features obtained by causality analysis were introduced to depict the interaction information among a set of objects. In order to cope with feature noisy and uncertainty, a method for multiple-object anomaly detection was presented via a sparse reconstruction. The abnormality of the testing sample was decided by the sparse reconstruction cost from an atomically learned dictionary. Experiment results show the effectiveness of the proposed method in comparison with other state-of-the-art methods on the public databases for abnormal behavior detection.
文摘In this study,we propose a low-cost system that can detect the space outlier utilization of residents in an indoor environment.We focus on the users’app usage to analyze unusual behavior,especially in indoor spaces.This is reflected in the behavioral analysis in that the frequency of using smartphones in personal spaces has recently increased.Our system facilitates autonomous data collection from mobile app logs and Google app servers and generates a high-dimensional dataset that can detect outlier behaviors.The density-based spatial clustering of applications with noise(DBSCAN)algorithm was applied for effective singular movement analysis.To analyze high-level mobile phone usage,the t-distributed stochastic neighbor embedding(t-SNE)algorithm was employed.These two clustering algorithms can effectively detect outlier behaviors in terms of movement and app usage in indoor spaces.The experimental results showed that our system enables effective spatial behavioral analysis at a low cost when applied to logs collected in actual living spaces.Moreover,large volumes of data required for outlier detection can be easily acquired.The system can automatically detect the unusual behavior of a user in an indoor space.In particular,this study aims to reflect the recent trend of the increasing use of smartphones in indoor spaces to the behavioral analysis.
文摘Detecting malware on mobile devices using the Android operating system has become a critical challenge in the field of cybersecurity,in the context of the rapid increase in the number of malware variants and the frequency of attacks targeting Android devices.In this paper,we propose a novel intelligent computational method to enhance the effectiveness of Android malware detection models.The proposed method combines two main techniques:(1)constructing a malware behavior profile and(2)extracting features from the malware behavior profile using graph neural networks.Specifically,to effectively construct an Android malware behavior profile,this paper proposes an information enrichment technique for the function call graph of malware files,based on new graph-structured features and semantic features of the malware’s source code.Additionally,to extract significant features from the constructed behavior profile,the study proposes using the GraphSAGE graph neural network.With this novel intelligent computational method,a variety of significant features of the malware have been effectively represented,synthesized,and extracted.The approach to detecting Android malware proposed in this paper is a new study and has not been explored in previous research.The experimental results on a dataset of 40,819 Android software indicate that the proposed method performs well across all metrics,with particularly impressive accuracy and recall scores of 99.03%and 99.19%,respectively,which outperforms existing state-of-the-art methods.
基金National Natural Science Foundation of China(61701029)。
文摘To improve the detection accuracy and robustness of crowd anomaly detection,especially crowd emergency evacuation detection,the abnormal crowd behavior detection method is proposed.This method is based on the improved statistical global optical flow entropy which can better describe the degree of chaos of crowd.First,the optical flow field is extracted from the video sequences and a 2D optical flow histogram is gained.Then,the improved optical flow entropy,combining information theory with statistical physics is calculated from 2D optical flow histograms.Finally,the anomaly can be detected according to the abnormality judgment formula.The experimental results show that the detection accuracy achieved over 95%in three public video datasets,which indicates that the proposed algorithm outperforms other state-of-the-art algorithms.
文摘Abnormal behavior detection is challenging and one of the growing research areas in computer vision.The main aim of this research work is to focus on panic and escape behavior detections that occur during unexpected/uncertain events.In this work,Pyramidal Lucas Kanade algorithm is optimized using EME-HOs to achieve the objective.First stage,OPLKT-EMEHOs algorithm is used to generate the opticalflow from MIIs.Second stage,the MIIs opticalflow is applied as input to 3 layer CNN for detect the abnormal crowd behavior.University of Minnesota(UMN)dataset is used to evaluate the proposed system.The experi-mental result shows that the proposed method provides better classification accu-racy by comparing with the existing methods.Proposed method provides 95.78%of precision,90.67%of recall,93.09%of f-measure and accuracy with 91.67%.
基金supported by the Nuclear Safety Research Program through Korea Foundation of Nuclear Safety(KoFONS)using the financial resource granted by the Nuclear Safety and Security Commission(NSSC)of the Republic of Korea(Grant number:2106061,50%)supported by the Institute of Information&Communications Technology Planning&Evaluation(IITP)grant funded by the Korea government(MSIT)(RS-2025-25394739,Development of Security Enhancement Technology for Industrial Control Systems Based on S/HBOM Supply Chain Protection,50%).
文摘With the continuous expansion of digital infrastructures,malicious behaviors in host systems have become increasingly sophisticated,often spanning multiple processes and employing obfuscation techniques to evade detection.Audit logs,such as Sysmon,offer valuable insights;however,existing approaches typically flatten event sequences or rely on generic graph models,thereby discarding the natural parent-child process hierarchy that is critical for analyzing multiprocess attacks.This paper proposes a structure-aware threat detection framework that transforms audit logs into a unified two-dimensional(2D)spatio-temporal representation,where process hierarchy is modeled as the spatial axis and event chronology as the temporal axis.In addition,entropy-based features are incorporated to robustly capture obfuscated and non-linguistic strings,overcoming the limitations of semantic embeddings.The model’s performance was evaluated on publicly available datasets,achieving competitive results with an accuracy exceeding 95%and an F1-score of at least 0.94.The proposed approach provides a promising and reproducible solution for detecting attacks with unknown indicators of compromise(IoCs)by analyzing the relationships and behaviors of processes recorded in large-scale audit logs.
文摘With technology constantly becoming present in people’s lives, smart homes are increasing in popularity. A smart home system controls lighting, temperature, security camera systems, and appliances. These devices and sensors are connected to the internet, and these devices can easily become the target of attacks. To mitigate the risk of using smart home devices, the security and privacy thereof must be artificially smart so they can adapt based on user behavior and environments. The security and privacy systems must accurately analyze all actions and predict future actions to protect the smart home system. We propose a Hybrid Intrusion Detection (HID) system using machine learning algorithms, including random forest, X gboost, decision tree, K -nearest neighbors, and misuse detection technique.
基金supported by the Fundamental Research Funds for the Central Universities(No.NJ20160015)
文摘In recent years,unmanned air vehicles(UAVs)are widely used in many military and civilian applications.With the big amount of UAVs operation in air space,the potential security and privacy problems are arising.This can lead to consequent harm for critical infrastructure in the event of these UAVs being used for criminal or terrorist purposes.Therefore,it is crucial to promptly identify the suspicious behaviors from the surrounding UAVs for some important regions.In this paper,a novel fuzzy logic based UAV behavior detection system has been presented to detect the different levels of risky behaviors of the incoming UAVs.The heading velocity and region type are two input indicators proposed for the risk indicator output in the designed fuzzy logic based system.The simulation has shown the effective and feasible of the proposed algorithm in terms of recall and precision of the detection.Especially,the suspicious behavior detection algorithm can provide a recall of 0.89 and a precision of 0.95 for the high risk scenario in the simulation.
基金supported by the Major Project of the National Natural Science Foundation of China (82090051,81871442)Outstanding Member Project of Youth Innovation Promotion Association of the Chinese Academy of Sciences (Y201930)。
文摘Quantification of behaviors in macaques provides crucial support for various scientific disciplines,including pharmacology,neuroscience,and ethology.Despite recent advancements in the analysis of macaque behavior,research on multi-label behavior detection in socially housed macaques,including consideration of interactions among them,remains scarce.Given the lack of relevant approaches and datasets,we developed the Behavior-Aware Relation Network(BARN)for multi-label behavior detection of socially housed macaques.Our approach models the relationship of behavioral similarity between macaques,guided by a behavior-aware module and novel behavior classifier,which is suitable for multi-label classification.We also constructed a behavior dataset of rhesus macaques using ordinary RGB cameras mounted outside their cages.The dataset included 65?913 labels for19 behaviors and 60?367 proposals,including identities and locations of the macaques.Experimental results showed that BARN significantly improved the baseline SlowFast network and outperformed existing relation networks.In conclusion,we successfully achieved multilabel behavior detection of socially housed macaques with both economic efficiency and high accuracy.
