检索结果-内蒙古大学图书馆

29th International Conference on Artificial Neural Networks (ICANN)

作者： Hahner, Sara Iza-Teran, Rodrigo Garcke, Jochen Fraunhofer Ctr Machine Learning St Augustin Germany SCAI St Augustin Germany Univ Bonn Inst Numer Simulat Bonn Germany

ISBN: (纸本)9783030616090;9783030616083

For sequences of complex 3D shapes in time we present a general approach to detect patterns for their analysis and to predict the deformation by making use of structural components of the complex shape. We incorporate long short-term memory (lstm) layers into an autoencoder to create low dimensional representations that allow the detection of patterns in the data and additionally detect the temporal dynamics in the deformation behavior. This is achieved with two decoders, one for reconstruction and one for prediction of future time steps of the sequence. In a preprocessing step the components of the studied object are converted to oriented bounding boxes which capture the impact of plastic deformation and allow reducing the dimensionality of the data describing the structure. The architecture is tested on the results of 196 car crash simulations of a model with 133 different components, where material properties are varied. In the latent representation we can detect patterns in the plastic deformation for the different components. The predicted bounding boxes give an estimate of the final simulation result and their quality is improved in comparison to different baselines.

关键词： lstm autoencoder 3D shape deformation analysis CAE analysis 3D time series

来源：评论

学校读者我要写书评

暂无评论

Detecting APS failures using lstm-AE and anomaly transformer enhanced with human expert analysis

引用

ENGINEERING FAILURE ANALYSIS 2024年 165卷

作者： Mumcuoglu, Mehmet E. Farea, Shawqi M. Unel, Mustafa Mise, Serdar Unsal, Simge Cevik, Enes Yilmaz, Metin Koprubasi, Kerem Sabanci Univ Fac Engn & Nat Sci TR-34956 Istanbul Turkiye Ford Otosan R&D Prod Dev Team TR-34885 Istanbul Turkiye

This study develops a novel semi-supervised approach for detecting Air Pressure System (APS) failures in Heavy-Duty Vehicles (HDVs) by exploiting two modern Machine Learning (ML) models: Long Short-Term Memory autoencoder (lstm-AE) and Transformer for Anomaly Detection (TranAD), and enhancing their performance with human expertise. To tackle the failure detection problem, a dataset comprising 30 days of operational time-series data from 110 healthy vehicles with no recorded APS issues and 30 vehicles that experienced APS failures requiring road assistance was acquired. Several preprocessing steps are proposed and three key features are extracted as APS health indicators. These features are then utilized both in human expert analysis (HEA) and training of ML models. When compared to HEA, both lstm-AE and TranAD models exhibit superior performance individually in APS failure detection, achieving F1 scores of 0.75 and 0.79 respectively, and the same accuracy of 91.4%. Further, the integration of HEA with those ML models enhances model effectiveness in all experimental results, especially in reducing false alarms that cause customer dissatisfaction. The TranAD model combined with human expert analysis achieved the best performance with an unprecedented 0.82 F1 score and 92.8% accuracy. In addition to presenting a new methodology for failure detection, this paper suggests a way for more efficient and reliable predictive maintenance practices for HDVs.

关键词： Heavy-duty vehicles Air pressure system Failure detection Predictive maintenance TranAD lstm autoencoder

来源：评论

学校读者我要写书评

暂无评论

TimeTector: A Twin-Branch Approach for Unsupervised Anomaly Detection in Livestock Sensor Noisy Data (TT-TBAD)

引用

SENSORS 2024年第8期24卷 2453页

作者： Kakar, Junaid Khan Hussain, Shahid Kim, Sang Cheol Kim, Hyongsuk Jeonbuk Natl Univ Dept Elect & Informat Engn Jeonju 54896 South Korea Jeonbuk Natl Univ Core Res Inst Intelligent Robots Jeonju 54896 South Korea Natl Univ Ireland Maynooth NUIM Innovat Value Inst IVI Sch Business Maynooth W23 F2H6 Ireland Jeonbuk Natl Univ Dept Elect Engn Jeonju 54896 South Korea

Unsupervised anomaly detection in multivariate time series sensor data is a complex task with diverse applications in different domains such as livestock farming and agriculture (LF&A), the Internet of Things (IoT), and human activity recognition (HAR). Advanced machine learning techniques are necessary to detect multi-sensor time series data anomalies. The primary focus of this research is to develop state-of-the-art machine learning methods for detecting anomalies in multi-sensor data. Time series sensors frequently produce multi-sensor data with anomalies, which makes it difficult to establish standard patterns that can capture spatial and temporal correlations. Our innovative approach enables the accurate identification of normal, abnormal, and noisy patterns, thus minimizing the risk of misinterpreting models when dealing with mixed noisy data during training. This can potentially result in the model deriving incorrect conclusions. To address these challenges, we propose a novel approach called "TimeTector-Twin-Branch Shared lstm autoencoder" which incorporates several Multi-Head Attention mechanisms. Additionally, our system now incorporates the Twin-Branch method which facilitates the simultaneous execution of multiple tasks, such as data reconstruction and prediction error, allowing for efficient multi-task learning. We also compare our proposed model to several benchmark anomaly detection models using our dataset, and the results show less error (MSE, MAE, and RMSE) in reconstruction and higher accuracy scores (precision, recall, and F1) against the baseline models, demonstrating that our approach outperforms these existing models.

关键词： anomaly detection unsupervised learning time series sensor lstm autoencoder multi-head attention mechanism DAE reconstruction and prediction branch livestock

来源：评论

学校读者我要写书评

暂无评论

Explainable Anomaly Detection for Industrial Control System Cybersecurity 10th

Explainable Anomaly Detection for Industrial Control System ...

引用

10th IFAC Triennial Conference on Manufacturing Modelling, Management and Control (MIM)

作者： Do Thu Ha Nguyen Xuan Hoang Nguyen Viet Hoang Nguyen Huu Du Truong Thu Huong Kim Phuc Tran Dong A Univ Int Res Inst Artificial Intelligence & Data Sci Danang Vietnam Hanoi Univ Sci & Technol Hanoi Vietnam Univ Lille ENSAIT ULR 2461 GEMTEX Genie & Mat Text F-59000 Lille France

Industrial Control Systems (ICSs) are becoming more and more important in managing the operation of many important systems in smart manufacturing, such as power stations, water supply systems, and manufacturing sites. While massive digital data can be a driving force for system performance, data security has raised serious concerns. Anomaly detection, therefore, is essential for preventing network security intrusions and system attacks. Many AI-based anomaly detection methods have been proposed and achieved high detection performance, however, are still a "black box" that is hard to be interpreted. In this study, we suggest using Explainable Artificial Intelligence to enhance the perspective and reliable results of an lstm-based autoencoder-OCSVM learning model for anomaly detection in ICS. We demonstrate the performance of our proposed method based on a well-known SCADA dataset. Copyright (C) 2022 The Authors.

关键词： XAI lstm autoencoder Anomaly Detection ICS Gradient SHAP

来源：评论

学校读者我要写书评

暂无评论

A Manipulated Overlapped Voltage Attack Detection Mechanism for Voltage-Based Vehicle Intrusion Detection System 1

引用

5th International Conference on Frontiers in Cyber Security (FCS)

作者： Yin, Long Xu, Jian Chai, Heqiu Wang, Chen Northeastern Univ Software Coll Shenyang 110169 Peoples R China

ISBN: (数字)9789811984457

ISBN: (纸本)9789811984440;9789811984457

To evade being detected by the content-based or frequency-based IDS, the attack model in the automotive CAN has shifted from the traditional packet flooding and payload modification attacks to stealth attacks such as shutdown attacks. These new types of stealth attacks are difficult to be effectively detected by content-based IDS and frequency-based IDS. The CAN bus physical voltage-based IDS can identify the source of each message and detect these stealth attacks effectively. However, the state of art research has discovered a novel masquerade attack called DUET, which can tamper with the existing voltage-based IDS by generating overlapping voltage signals with an accomplice to distort the fingerprint of the specified ECU. We propose a detection mechanism to prevent the manipulated voltage attacks of overlapping voltage signal samples, which is based on anomaly detection by applying the lstm autoencoder model. By filtering the overlapped signal and rectifying the voltage fingerprint instance of the original voltage signal, the improved voltage-based IDS can effectively resist the DUET attack. Experiments demonstrated the proposed detection mechanism can authenticate the victim ECU and the accomplice ECU before and after the DUET two-stage attack, and prevent the receiver ECU from being deceived by the forged messages generated by the attacker and accomplice ECUs.

关键词： Voltage-based IDS DUET attack Manipulated overlapped voltage attack lstm autoencoder

来源：评论

学校读者我要写书评

暂无评论

Revisiting the Non-recurrent Traffic Incident Identification Problem for Real-Time Applications Using Theory-Aware Unsupervised Learning

引用

Data Science for Transportation 2025年第2期7卷 1-17页

作者： Papadatou, Katerina Ν. Fafoutellis, Panagiotis Vlahogianni, Eleni I. Department of Transportation Planning and Engineering School of Civil Engineering National Technical University of Athens Athens Greece

In this paper, we present a novel approach for real-time detection of non-recurrent traffic patterns in urban roadway networks leveraging advanced machine learning techniques explained by traffic flow theory. The methodology comprises two key components. First, an lstm-based autoencoder is employed to extract typical expected traffic patterns from raw traffic data. Second, a clustering technique is utilized to identify non-recurrent congestion, applied on the deviation of the speed and volume measurements from the aforementioned typical patterns. The methodology is implemented using high-resolution multimodal traffic data from the urban road network of Athens, Greece. Findings reveal the presence of four distinct traffic states, three of which represent various types of non-recurrent traffic conditions, characterized by their deviations from typical traffic patterns. The methodology can promptly detect the shift between recurrent and non-recurrent conditions in real time and may have far reaching implications for efficient urban traffic management systems.

关键词： Incident detection lstm autoencoder Non-recurrent traffic patterns Traffic conditions taxonomy Unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

Objective Falls Risk Assessment Using Markerless Motion Capture and Representational Machine Learning

引用

SENSORS 2024年第14期24卷 4593页

作者： Maudsley-Barton, Sean Yap, Moi Hoon Manchester Metropolitan Univ Dept Comp & Math Manchester M15 6BH England

Falls are a major issue for those over the age of 65 years worldwide. Objective assessment of fall risk is rare in clinical practice. The most common methods of assessment are time-consuming observational tests (clinical tests). Computer-aided diagnosis could be a great help. A popular clinical test for fall risk is the five times sit-to-stand. The time taken to complete the test is the most commonly used metric to identify the most at-risk patients. However, tracking the movement of skeletal joints can provide much richer insights. We use markerless motion capture, allied with a representational model, to identify those at risk of falls. Our method uses an lstm autoencoder to derive a distance measure. Using this measure, we introduce a new scoring system, allowing individuals with differing falls risks to be placed on a continuous scale. Evaluating our method on the KINECAL dataset, we achieved an accuracy of 0.84 in identifying those at elevated falls risk. In addition to identifying potential fallers, our method could find applications in rehabilitation. This aligns with the goals of the KINECAL Dataset. KINECAL contains the recordings of 90 individuals undertaking 11 movements used in clinical assessments. KINECAL is labelled to disambiguate age-related decline and falls risk.

关键词： anomaly detection computer-aided diagnosis falls risk lstm autoencoder representational model markerless motion capture

来源：评论

学校读者我要写书评

暂无评论

Driving Maneuver Classification Using Domain Specific Knowledge and Transfer Learning

引用

IEEE ACCESS 2021年 9卷 86590-86606页

作者： Sarker, Supriya Haque, Md Mokammel Dewan, M. Ali Akber Chittagong Univ Engn & Technol Dept Comp Sci & Engn Chattogram 4349 Bangladesh Athabasca Univ Fac Sci & Technol Sch Comp & Informat Syst Athabasca AB T9S 3A3 Canada

With the increasing number of vehicles, the usage of technology has also been increased in the transportation system. Although automobile companies are using advanced technologies to develop high performing transports, traffic safety still remains to be a concerning issue. Drivers' driving behavior is considered as one of the key factors of the traffic safety, which could be monitored from their individual driving maneuvers. In this paper, we present a supervised learning model and a semi-supervised transfer learning model for the classification of driving maneuvers from the sensor fusion time series data. The semi-supervised model consists of an unsupervised long-short term memory (lstm) autoencoder and a supervised lstm classifier. The supervised model consists of a supervised lstm model. Because of using lstm, both of the models can analyze time-series data. In the semi-supervised model, the lstm encoder learns from unlabeled data as a compressed low dimensional feature vector, which then transfers the learning to the supervised lstm classifier to classify the driving maneuvers. With the proposed models, we use domain specific knowledge data of the driving environment, such as data changing rules of various driving maneuvers as well as the temporal features over time. We use class functions for seven driving maneuver types and convert those into binary feature vector to use with the lstm models. We present a comparative analysis of the per class accuracy of the proposed semi-supervised and supervised models with and without using domain-specific knowledge, where the models with the domain specific knowledge outperform. Our proposed semi-supervised and supervised models are compared with the other existing approaches, where our models trained with the domain specific knowledge provided better performance. We also compared the per class accuracy for both the supervised and semi-supervised models, where all the maneuver class accuracy for supervised model was above

关键词： Data models Accelerometers Vehicles Smart phones Time series analysis Safety Gyroscopes Driving maneuver classification domain specific knowledge lstm autoencoder semi-supervised learning transfer learning

来源：评论

学校读者我要写书评

暂无评论

Dynamic Facial Expression Understanding Using Deep Spatiotemporal LDSP On Spark

引用

IEEE ACCESS 2021年 9卷 16866-16877页

作者： Uddin, Md Azher Joolee, Joolekha Bibi Sohn, Kyung-Ah Ajou Univ Dept Artificial Intelligence Suwon 16499 South Korea Kyung Hee Univ Dept Comp Sci & Engn Global Campus Yongin 17104 South Korea Ajou Univ Dept Software & Comp Engn Suwon 16499 South Korea

Facial expressions are the most common medium for expressing human emotions. Due to the wide range of real-world applications, facial expression understanding has received extensive attention from researchers. One of the most vital issues of facial expression recognition is the extraction and modeling of the temporal dynamics of facial emotions from videos. Additionally, the rapid growth of video data from various multimedia sources is becoming a serious concern. Therefore, to address these issues, in this paper, we introduce a novel approach on top of Spark for facial expression understanding from videos. First, we propose a new dynamic feature descriptor, namely, the local directional structural pattern from three orthogonal planes (LDSP-TOP), which analyzes the structural aspects of the local dynamic texture. Second, we design a 1-D convolutional neural network (CNN) to capture additional discriminative features. Third, a long short-term memory (lstm) autoencoder is employed to learn the spatiotemporal features. Finally, an extensive experimental investigation is carried out to demonstrate the performance and scalability of the proposed framework.

关键词： Facial expression understanding local directional structural pattern from three orthogonal planes 1-D CNN lstm autoencoder

来源：评论

学校读者我要写书评

暂无评论

Edge AI for Real-Time Anomaly Detection in Smart Homes

引用

FUTURE INTERNET 2025年第4期17卷 179-179页

作者： Reis, Manuel J. C. S. Serodio, Carlos Univ Tras Os Montes & Alto Douro Engn Dept P-5000801 Vila Real Portugal IEETA Inst Elect & Informat Engn Aveiro P-3810193 Aveiro Portugal Algoritmi Ctr P-4800058 Guimaraes Portugal

The increasing adoption of smart home technologies has intensified the demand for real-time anomaly detection to improve security, energy efficiency, and device reliability. Traditional cloud-based approaches introduce latency, privacy concerns, and network dependency, making Edge AI a compelling alternative for low-latency, on-device processing. This paper presents an Edge AI-based anomaly detection framework that combines Isolation Forest (IF) and Long Short-Term Memory autoencoder (lstm-AE) models to identify anomalies in IoT sensor data. The system is evaluated on both synthetic and real-world smart home datasets, including temperature, motion, and energy consumption signals. Experimental results show that lstm-AE achieves higher detection accuracy (up to 93.6%) and recall but requires more computational resources. In contrast, IF offers faster inference and lower power consumption, making it suitable for constrained environments. A hybrid architecture integrating both models is proposed to balance accuracy and efficiency, achieving sub-50 ms inference latency on embedded platforms such as Raspberry Pi and NVIDEA Jetson Nano. Optimization strategies such as quantization reduced lstm-AE inference time by 76% and power consumption by 35%. Adaptive learning mechanisms, including federated learning, are also explored to minimize cloud dependency and enhance data privacy. These findings demonstrate the feasibility of deploying real-time, privacy-preserving, and energy-efficient anomaly detection directly on edge devices. The proposed framework can be extended to other domains such as smart buildings and industrial IoT. Future work will investigate self-supervised learning, transformer-based detection, and deployment in real-world operational settings.

关键词： edge AI anomaly detection smart home IoT isolation forest lstm autoencoder

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

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

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：