Article
Computer Science, Artificial Intelligence
Jiawei Yang, Xu Tan, Sylwan Rahardja
Summary: Unsupervised k-nearest-neighbor-based outlier detectors are important in data science research. This article proposes a new concept - neighborhood consistency, to tackle the challenge of selecting the optimal k. The developed KFC method outperforms baselines and exhibits good generality.
PATTERN RECOGNITION LETTERS
(2023)
Article
Computer Science, Software Engineering
Muhammad Umer Farooq, Mohamad Naufal M. Saad, Sultan Daud Khan
Summary: A novel method for abnormal crowd behavior detection in surveillance videos is proposed, focusing on crowd divergence behavior using convolutional neural networks trained on motion-shape images. Experimental results demonstrate the method's robustness and superior accuracy compared to existing methods.
Article
Computer Science, Artificial Intelligence
Jacinto Carrasco, David Lopez, Ignacio Aguilera-Martos, Diego Garcia-Gil, Irina Markova, Marta Garcia-Barzana, Manuel Arias-Rodil, Julian Luengo, Francisco Herrera
Summary: The research in anomaly detection lacks a unified definition of what represents an anomalous instance, leading to diverse paradigms in algorithm design and experimentation. Predictive maintenance represents a special case, where anomalies must be prevented to avoid failures. To address these issues, the concept of positive and negative instances has been generalized into intervals for evaluating unsupervised anomaly detection algorithms.
Article
Computer Science, Artificial Intelligence
Pengyun Zhu, Chaowei Zhang, Xiaofeng Li, Jifu Zhang, Xiao Qin
Summary: Traditional outlier detection methods are not suitable for high-dimensional data analysis due to the curse of dimensionality. Inspired by Coulomb's law, a new similarity measure vector is proposed for high-dimensional data, which incorporates outlier Coulomb force and outlier Coulomb resultant force. The algorithm effectively measures similarity and differences among data objects, and provides interpretable results with the Coulomb resultant force. The algorithm is evaluated using UCI and synthetic datasets, demonstrating its effectiveness in alleviating the curse of dimensionality and producing interpretable high-dimensional outlier data.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Madalina Olteanu, Fabrice Rossi, Florian Yger
Summary: The impact of outliers and anomalies on model estimation and data processing is highly significant, as shown by the extensive research in various fields over several decades. However, organizing and summarizing this research is challenging due to the pervasive nature of outliers and anomalies in all data-intensive applications. Given this need, this paper conducts a systematic meta-survey of general surveys and reviews on outlier and anomaly detection to provide insights into the evolution of this field and the writing practices of researchers.
Article
Computer Science, Artificial Intelligence
Jiawei Yang, Susanto Rahardja, Pasi Franti
Summary: The mean-shift outlier detector modifies data using mean-shift technique to eliminate the bias caused by outliers and remove their influence without needing to know the outliers. Experimental results show that this method performs well regardless of the number of outliers in the data.
PATTERN RECOGNITION
(2021)
Article
Physics, Multidisciplinary
Michiel Nijhuis, Iman van Lelyveld
Summary: Outliers are commonly found in data, and various algorithms exist to detect them. The verification of these outliers can determine whether they are data errors or not. However, this verification process is time-consuming and the underlying issues leading to the data error can change over time. Therefore, using reinforcement learning on a statistical outlier detection approach can optimize the detection process by adjusting the coefficients of the ensemble model with every new piece of data.
Article
Computer Science, Artificial Intelligence
Luis Antonio Souto Arias, Cornelis W. Oosterlee, Pasquale Cirillo
Summary: Many unsupervised anomaly detection algorithms rely on the concept of nearest neighbours to compute the anomaly scores. However, the number of nearest neighbours cannot be tuned in an unsupervised setting, so we propose the new Analytic Isolation and Distance-based Anomaly (AIDA) detection algorithm that combines distance metrics with isolation. We also introduce the Tempered Isolation-based eXplanation (TIX) algorithm, which improves the explainability of outlier detection in large multi-dimensional datasets.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Information Systems
Feng Guo, Fumin Zou, Sijie Luo, Lyuchao Liao, Jinshan Wu, Xiang Yu, Cheng Zhang
Summary: China's expressway electronic toll collection system generates a large amount of transaction data and records the traffic trajectories of almost all vehicles. However, there are false and missing rates in the data due to missed and false transactions. This paper proposes an improved DTW algorithm and a dynamic search step algorithm to improve the efficiency of abnormal data detection, providing a feasible method for detecting abnormal events in expressway ETC data.
Article
Computer Science, Artificial Intelligence
Felix Iglesias Vazquez, Alexander Hartl, Tanja Zseby, Arthur Zimek
Summary: This study evaluates and compares eight state-of-the-art algorithms for anomaly detection in streaming data, taking into account the challenges posed by computational costs and nonstationarity. The results shed light on the importance of considering factors such as locality, relativeness, and concept drift in order to select the most appropriate algorithm. By leveraging historical data and domain knowledge, optimal designs for streaming data anomaly detection in real-life applications can be achieved.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Zhong Li, Matthijs van Leeuwen
Summary: Traditional anomaly detection methods treat all features equally to identify objects that deviate from most other objects. In contrast, contextual anomaly detection methods divide the features into contextual and behavioral features to detect objects that deviate from other objects within a context of similar objects. This paper establishes connections between dependency-based traditional anomaly detection methods and contextual anomaly detection methods, and proposes a novel approach to inherently interpretable contextual anomaly detection using Quantile Regression Forests to model dependencies between features. Extensive experiments on various synthetic and real-world datasets demonstrate that our method outperforms state-of-the-art anomaly detection methods in terms of accuracy and interpretability.
DATA MINING AND KNOWLEDGE DISCOVERY
(2023)
Article
Computer Science, Artificial Intelligence
Byung Il Kwak, Mee Lan Han, Huy Kang Kim
Summary: In recent years, the advancement of vehicular technology and the increasing connectivity between vehicles and the external environment have highlighted the importance of addressing security issues. This study proposes an anomaly detection method based on cosine similarity for in-vehicle networks to detect different types of injection attacks effectively.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Automation & Control Systems
Oded Koren, Michal Koren, Or Peretz
Summary: Anomaly detection is important for identifying and removing outliers in datasets. This paper presents a feature-wise anomaly detection procedure that uses different techniques and a new measure called Noise Ratio. Results show that the proposed method outperforms traditional anomaly detection algorithms in terms of detecting outliers.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Mathematics
Jorge R. Sosa Donoso, Miguel Flores, Salvador Naya, Javier Tarrio-Saavedra
Summary: This work presents a methodology for detecting outliers in functional data that considers both their shape and magnitude. The Local Correlation Integral (LOCI) method, a multivariate anomaly detection technique, has been extended and adapted for functional data using distance calculations in Hilbert spaces. The methodology has been validated through simulation studies and application to real data, showing good performance in scenarios with inter-curve dependence, particularly when outliers are due to curve magnitudes. Results are further supported by the successful application of the methodology to a meteorological database, outperforming other competitive methods.
Article
Computer Science, Information Systems
Xusheng Du, Jiong Yu, Zheng Chu, Lina Jin, Jiaying Chen
Summary: Outlier detection technologies are crucial in various domains. The study proposes a novel graph neural network structure called the graph autoencoder (GAE) for outlier detection in Euclidean structured data. Experimental results show that GAE achieves the highest ROC AUC among multiple datasets.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Wenshan Wang, Su Yang, Weishan Zhang
Summary: In this article, we explore three dynamic weighting ensemble learning models to fuse spatio-temporal features and visual features for customer volume prediction in urban commercial districts. We introduce the shared-private dynamic weighting model with graph neural networks to capture geographic dependencies between districts. The effectiveness of the proposed models is demonstrated through experiments on real datasets, and a visualization method is used for knowledge discovery.
ACM TRANSACTIONS ON INTELLIGENT SYSTEMS AND TECHNOLOGY
(2023)
Article
Computer Science, Information Systems
Junsan Zhang, Xiuxuan Shen, Shaohua Wan, Sotirios K. Goudos, Jie Wu, Ming Cheng, Weishan Zhang
Summary: Automatic generation of medical reports can assist doctors and reduce their workload. However, previous methods of injecting auxiliary information had limitations, which are addressed by the proposed Information Calibrated Transformer (ICT) that extracts features from datasets as auxiliary information and combines them with the main model through an Information Calibration Attention Module (ICA), resulting in improved quality of generated medical reports.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2023)
Article
Computer Science, Information Systems
Shaohua Cao, Shu Chen, Hui Chen, Hanqing Zhang, Zijun Zhan, Weishan Zhang
Summary: With the growth of the Internet of Things, mobile edge computing is proposed as a solution to reduce service latency and congestion in mobile core networks. This paper presents a hybrid computation offloading framework based on Deep Deterministic Policy Gradient (DDPG) in IoT. The framework consists of edge servers and user devices, and uses Software Defined Network technologies to acquire the environment state and generate offloading strategies. Experimental results show that the proposed algorithm outperforms the comparison algorithm and improves system latency and stability of network and computational load.
Article
Computer Science, Cybernetics
Tao Chen, Baoyu Zhang, Xiao Wang, Weishan Zhang, Chitin Hon, Wang Di, Long Chen, Qiang Li, Fei-Yue Wang
Summary: The dynamics of public opinion on social media affect people's perceptions of international affairs and lead to the restructuring of social states in international conflicts. This article analyzes the evolution of topics on social media during Pelosi's visit. Such analysis helps related departments effectively sense and respond to the situation, and provides technical support for policy making. A new method and ALGC algorithm are proposed to reduce the computational complexity of large graphs and analyze the evolution pattern of topics. The experimental results show that the proposed method achieves high clustering accuracy with lower computational cost. The dataset used in this article is also released for public analysis.
IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS
(2023)
Article
Computer Science, Hardware & Architecture
Zhicheng Bao, Weishan Zhang, Xingjie Zeng, Hongwei Zhao, Cihao Dong, Yuming Nie, Yuru Liu, Yuange Liu, Junzhong Wu
Summary: In this article, a comprehensive approach to a responsible AI-based software architecture for the digitalization of industry drawings is proposed, which also serves as a software engineering reference for responsible AI in other industry domains.
Article
Computer Science, Hardware & Architecture
Shaohua Cao, Di Liu, Congcong Dai, Chengqi Wang, Yansheng Yang, Weishan Zhang, Danyang Zheng
Summary: With the development of autonomous and intelligent techniques, vehicles are equipped with computation and communication modules to handle on-vehicle computing requests. However, due to limited computation capacities, these requests are offloaded to special devices like roadside units or intelligent vehicles. Two challenges arise in vehicular edge computing networks: accurately determining peak or low hours and effectively offloading requests. This paper investigates computational requests offloading in different vehicular networking scenarios and proposes algorithms based on fuzzy inference and reinforcement learning to address these challenges. Experimental results show significant improvement in resource utilization compared to the benchmark.
Article
Computer Science, Information Systems
Xingjie Zeng, Zepei Yu, Weishan Zhang, Xiao Wang, Qinghua Lu, Tao Wang, Mu Gu, Yonglin Tian, Fei-Yue Wang
Summary: This article introduces a homophily learning-based federated intelligence (HLFI) approach, which uses hierarchical federated learning strategy and dynamic elimination learning strategy to address the issues in deep neural networks in federated learning. The experiments show that the proposed approach can improve equipment failure prediction F1-score up to 9.32% and has good generalization capabilities to improve model performance in other federated learning methods.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Automation & Control Systems
Weishan Zhang, Fa Yu, Xiao Wang, Xingjie Zeng, Hongwei Zhao, Yonglin Tian, Fei-Yue Wang, Longfei Li, Zengxiang Li
Summary: Resilient Reinforcement Federated Learning (R(2)Fed) is a method that applies reinforcement learning to federated learning and uses reinforcement learning for weighted fusion of client models, instead of average fusion, to tackle the impact of device heterogeneity and non-identically and independently distributed data (Non-IID) in actual federated learning. Experiments demonstrate that the R(2)Fed method outperforms traditional federated learning in Non-IID and heterogeneity scenarios, increasing the average accuracy by 4.7%. The experiments also show that R(2)Fed is resilient to federation attacks.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Computer Science, Information Systems
Liang Xu, Haoyun Sun, Hongwei Zhao, Weishan Zhang, Huansheng Ning, Hongqing Guan
Summary: This article proposes an accurate and efficient federated learning-based edge intelligence for effective video analysis method called EIEVA-AEFL. It includes a FMRN network to reduce object misdetection and an efficient federated learning strategy to reduce communication cost. Experimental results on datasets show that EIEVA-AEFL improves model accuracy and reduces training time.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Computer Science, Cybernetics
Liang Xu, Tao Chen, Zhaoxiang Hou, Weishan Zhang, Chitin Hon, Xiao Wang, Di Wang, Long Chen, Wenyin Zhu, Yunlong Tian, Huansheng Ning, Fei-Yue Wang
Summary: This article proposes a knowledge graph-based reinforcement federated learning (KGRFL)-based Q & A approach to address the challenges of error accumulation and data privacy in existing semantic parsing methods. By designing a multitask semantic parsing model and a reinforcement learning-based model fusion strategy, multi-institution joint modeling and data privacy protection are achieved. The experiments show that the proposed method outperforms using a single institution and is resilient to security attacks.
IEEE TRANSACTIONS ON COMPUTATIONAL SOCIAL SYSTEMS
(2023)
Article
Telecommunications
Xin Liu, Liang Zheng, Sumi Helal, Weishan Zhang, Chunfu Jia, Jiehan Zhou
Summary: This paper proposes a comprehensive defense approach against SSDP reflection attacks, combining broad learning and a set of defense strategies. It does not require detecting attacks or identifying the roles of IoT devices, and can accurately detect suspicious traffic, automatically deploy defense strategies, and significantly reduce DDoS packets.
DIGITAL COMMUNICATIONS AND NETWORKS
(2023)
Article
Telecommunications
Weishan Zhang, Xiao Chen, Ke He, Leiming Chen, Liang Xu, Xiao Wang, Su Yang
Summary: This paper introduces an efficient semi-asynchronous federated learning framework for short-term solar power forecasting, and evaluates the performance using a CNN-LSTM model. The research demonstrates that federated models can achieve higher forecasting performance while protecting data privacy.
DIGITAL COMMUNICATIONS AND NETWORKS
(2023)
Article
Engineering, Electrical & Electronic
Yuange Liu, Zhicheng Bao, Yuqian Wang, Xingjie Zeng, Liang Xu, Weishan Zhang, Hongwei Zhao, Zepei Yu
Summary: This paper proposes an adaptive personalized federated meta-learning framework to address the challenges in industrial equipment anomaly detection, with excellent ability to adjust hyperparameters and applicability to different environments.
IEEE JOURNAL OF RADIO FREQUENCY IDENTIFICATION
(2022)
Article
Computer Science, Information Systems
Huaqiang Kang, Yan Liu, Weishan Zhang
Summary: This article introduces a distributed trajectory segmentation framework, which includes a greedy-split segmentation method and a distributed spatial R-tree index of trajectory segments. The framework can identify moving groups of trajectories by defining metrics to measure trajectory similarity and chance of collision. The evaluation results show that the system performs well in terms of data partition, parallelism, and data size.
IEEE TRANSACTIONS ON CLOUD COMPUTING
(2022)
Article
Computer Science, Information Systems
Fadi Farha, Huansheng Ning, Shunkun Yang, Jiabo Xu, Weishan Zhang, Kim-Kwang Raymond Choo
Summary: ZigBee is a communication protocol used in IoT applications, but its security is compromised in typical low-cost and low-power deployment scenarios. This paper presents a timestamp-based scheme to mitigate replay attacks, which is power-efficient and applicable to different ZigBee topologies and end devices.
IEEE TRANSACTIONS ON MOBILE COMPUTING
(2022)
Article
Computer Science, Artificial Intelligence
Guang-Hai Liu, Zuo-Yong Li, Jing-Yu Yang, David Zhang
Summary: This article introduces a novel image retrieval method that improves retrieval performance by using sublimated deep features. The method incorporates orientation-selective features and color perceptual features, effectively mimicking these mechanisms to provide a more discriminating representation.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Fengguang Peng, Zihan Ding, Ziming Chen, Gang Wang, Tianrui Hui, Si Liu, Hang Shi
Summary: RGB-Thermal (RGB-T) semantic segmentation is an emerging task that aims to improve the robustness of segmentation methods under extreme imaging conditions by using thermal infrared modality. The challenges of foreground-background distinguishment and complementary information mining are addressed by proposing a cross modulation process with two collaborative components. Experimental results show that the proposed method achieves state-of-the-art performances on current RGB-T segmentation benchmarks.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Baihong Han, Xiaoyan Jiang, Zhijun Fang, Hamido Fujita, Yongbin Gao
Summary: This paper proposes a novel automatic prompt generation method called F-SCP, which focuses on generating accurate prompts for low-accuracy classes and similar classes. Experimental results show that our approach outperforms state-of-the-art methods on six multi-domain datasets.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Huikai Liu, Ao Zhang, Wenqian Zhu, Bin Fu, Bingjian Ding, Shengwu Xiong
Summary: Adverse weather conditions present challenges for computer vision tasks, and image de-weathering is an important component of image restoration. This paper proposes a multi-patch skip-forward structure and a Residual Deformable Convolutional module to improve feature extraction and pixel-wise reconstruction.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Oliver M. Crook, Mihai Cucuringu, Tim Hurst, Carola-Bibiane Schonlieb, Matthew Thorpe, Konstantinos C. Zygalakis
Summary: The transportation LP distance (TLP) is a generalization of the Wasserstein WP distance that can be applied directly to color or multi-channelled images, as well as multivariate time-series. TLP interprets signals as functions, while WP interprets signals as measures. Although both distances are powerful tools in modeling data with spatial or temporal perturbations, their computational cost can be prohibitively high for moderate pattern recognition tasks. The linear Wasserstein distance offers a method for projecting signals into a Euclidean space, and in this study, we propose linear versions of the TLP distance (LTLP) that show significant improvement over the linear WP distance in signal processing tasks while being several orders of magnitude faster to compute than the TLP distance.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Haitao Tian, Shiru Qu, Pierre Payeur
Summary: This paper proposes a method of target-dependent classifier, which optimizes the joint hypothesis of domain adaptation into a target-dependent hypothesis that better fits with the target domain clusters through an unsupervised fine-tuning strategy and the concept of meta-learning. Experimental results demonstrate that this method outperforms existing techniques in synthetic-to-real adaptation and cross-city adaptation benchmarks.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Qingsen Yan, Axi Niu, Chaoqun Wang, Wei Dong, Marcin Wozniak, Yanning Zhang
Summary: Deep learning-based methods have achieved remarkable results in the field of super-resolution. However, the limitation of paired training image sets has led researchers to explore self-supervised learning. However, the assumption of inaccurate downscaling kernel functions often leads to degraded results. To address this issue, this paper introduces KGSR, a kernel-guided network that trains both upscaling and downscaling networks to generate high-quality high-resolution images even without knowing the actual downscaling process.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Yifan Chen, Xuelong Li
Summary: Gait recognition is a popular technology for identification due to its ability to capture gait features over long distances without cooperation. However, current methods face challenges as they use a single network to extract both temporal and spatial features. To solve this problem, we propose a two-branch network that focuses on spatial and temporal feature extraction separately. By combining these features, we can effectively learn the spatio-temporal information of gait sequences.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Wei Shi, Wentao Zhang, Wei-shi Zheng, Ruixuan Wang
Summary: This article proposes a simple yet effective visualization framework called PAMI, which does not require detailed model structure and parameters to obtain visualization results. It can be applied to various prediction tasks with different model backbones and input formats.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Xiaobo Hu, Jianbo Su, Jun Zhang
Summary: This paper reviews the latest technologies in pattern recognition, highlighting their instabilities and failures in practical applications. From a control perspective, the significance of disturbance rejection in pattern recognition is discussed, and the existing problems are summarized. Finally, potential solutions related to the application of compensation on features are discussed to emphasize future research directions.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Andres Felipe Posada-Moreno, Nikita Surya, Sebastian Trimpe
Summary: Convolutional neural networks are widely used in critical systems, and explainable artificial intelligence has proposed methods for generating high-level explanations. However, these methods lack the ability to determine the location of concepts. To address this, we propose a novel method for automatic concept extraction and localization based on pixel-wise aggregations, and validate it using synthetic datasets.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Peng Bao, Jianian Li, Rong Yan, Zhongyi Liu
Summary: In this paper, a novel Dynamic Graph Contrastive Learning framework, DyGCL, is proposed to capture the temporal consistency in dynamic graphs and achieve good performance in node representation learning.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Kristian Schultz, Saptarshi Bej, Waldemar Hahn, Markus Wolfien, Prashant Srivastava, Olaf Wolkenhauer
Summary: Research indicates that deep generative models perform poorly compared to linear interpolation-based methods for synthetic data generation on small, imbalanced tabular datasets. To address this, a new approach called ConvGeN, combining convex space learning with deep generative models, has been proposed. ConvGeN improves imbalanced classification on small datasets while remaining competitive with existing linear interpolation methods.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Khondaker Tasrif Noor, Antonio Robles-Kelly
Summary: In this paper, the authors propose H-CapsNet, a capsule network designed for hierarchical image classification. The network effectively captures hierarchical relationships using dedicated capsules for each class hierarchy. A modified hinge loss is utilized to enforce consistency among the involved hierarchies. Additionally, a strategy for dynamically adjusting training parameters is presented to achieve better balance between the class hierarchies. Experimental results demonstrate that H-CapsNet outperforms competing hierarchical classification networks.
PATTERN RECOGNITION
(2024)
Article
Computer Science, Artificial Intelligence
Lei Liu, Guorun Li, Yuefeng Du, Xiaoyu Li, Xiuheng Wu, Zhi Qiao, Tianyi Wang
Summary: This study proposes a new agricultural image segmentation model called CS-Net, which uses Simple-Attention Block and Simpleformer to improve accuracy and inference speed, and addresses the issue of performance collapse of Transformers in agricultural image processing.
PATTERN RECOGNITION
(2024)
