Search Results (262)

Electricity theft, emerging as one of the severe cyberattacks in smart grids, causes significant economic losses. Due to the powerful expressive ability of deep neural networks (DNN), supervised and unsupervised DNN-based electricity theft detection (ETD) schemes have experienced widespread deployment. However, existing works have the following weak points: Supervised DNN-based schemes require abundant labeled anomalous samples for training, and even worse, cannot detect unseen theft patterns. To avoid the extensively labor-consuming activity of labeling anomalous samples, unsupervised DNNs-based schemes aim to learn the normality of time-series and infer an anomaly score for each data instance, but they fail to capture periodic features effectively. To address these challenges, this paper proposes a novel periodicity-enhanced deep one-class classification framework (PE-DOCC) based on a periodicity-enhanced transformer encoder, named Periodicformer encoder. Specifically, within the encoder, a novel criss-cross periodic attention is proposed to capture both horizontal and vertical periodic features. The Periodicformer encoder is pre-trained by reconstructing partially masked input sequences, and the learned latent representations are then fed into a one-class classification for anomaly detection. Extensive experiments on real-world datasets demonstrate that our proposed PE-DOCC framework outperforms state-of-the-art unsupervised ETD methods. Full article

The detection of acoustic emission events from various failing mechanisms, such as plastic deformations, is a critical element in the monitoring and timely detection of structural failures in infrastructures. This study focuses on the detection of such failures in metal gates at rivers’ lifting dams aiming to increase the reliability of river transport compared to the current situation, thereby, increasing the resilience of transport corridors. During our study, we used lifting dams in both France and Italy where river transport is thriving. A methodology was developed, processing corresponding acoustic emission recordings originating from lifting dams’ metal gates, using advanced denoising—preprocessing, various decompositions, and spectral embeddings associated with various latest nonlinear processing clustering techniques—thus providing a detailed cluster label morphology and profile of water gates’ normal operating area. Latest machine learning outlier detection algorithms, like One-Class Support Vector Machine, Variational Auto-Encoder, and others, were incorporated, producing a vector of confidence on upcoming out-of-the-normal gate operation and failure prediction, achieving detection contrast enhancement on out-of-the-normal operation points up to 400%. Full article

The classification of the wireless propagation channel between Line-of-Sight (LOS) or Non-Line-of-Sight (NLOS) is useful in the operation of wireless communication systems. The research community has increasingly investigated the application of machine learning (ML) to LOS/NLOS classification and this paper is part of this trend, but not all the different aspects of ML have been analyzed. In the general ML domain, poisoning and adversarial attacks and the related mitigation techniques are an active area of research. Such attacks aim to hamper the ML classification process by poisoning the data set. Mitigation techniques are designed to counter this threat using different approaches. Poisoning attacks in LOS/NLOS classification have not received significant attention by the wireless communication community and this paper aims to address this gap by proposing the application of a specific mitigation technique based on outlier detection algorithms. The rationale is that poisoned samples can be identified as outliers from legitimate samples. In particular, the study described in this paper proposes a recent outlier detection algorithm, which has low computing complexity: the sparse data observers (SDOs) algorithm. The study proposes a comprehensive analysis of both conventional and novel types of attacks and related mitigation techniques based on outlier detection algorithms for UltraWideBand (UWB) channel classification. The proposed techniques are applied to two data sets: the public eWINE data set with seven different UWB LOS/NLOS different environments and a radar data set with the LOS/NLOS condition. The results show that the SDO algorithm outperforms other outlier detection algorithms for attack detection like the isolation forest (iForest) algorithm and the one-class support vector machine (OCSVM) in most of the scenarios and attacks, and it is quite competitive in the task of increasing the UWB LOS/NLOS classification accuracy through sanitation in comparison to the poisoned model. Full article

The widespread deployment of deep neural networks (DNNs) in critical real-time applications has spurred significant research into their security and robustness. A key vulnerability identified is that DNN decisions can be maliciously altered by introducing carefully crafted noise into the input data, leading to erroneous predictions. This is known as an adversarial attack. In this paper, we propose a novel detection framework leveraging segmentation masks and image segmentation techniques to identify adversarial attacks on DNNs, particularly in the context of autonomous driving systems. Our defense technique considers two levels of adversarial detection. The first level mainly detects adversarial inputs with large perturbations using the U-net model and one-class support vector machine (SVM). The second level of defense proposes a dynamic segmentation algorithm based on the k-means algorithm and a verifier model that controls the final prediction of the input image. To evaluate our approach, we comprehensively compare our method to the state-of-the-art feature squeeze method under a white-box attack, using eleven distinct adversarial attacks across three benchmark and heterogeneous data sets. The experimental results demonstrate the efficacy of our framework, achieving overall detection rates exceeding 96% across all adversarial techniques and data sets studied. It is worth mentioning that our method enhances the detection rates of FGSM and BIM attacks, reaching average detection rates of 95.65% as opposed to 62.63% in feature squeezing across the three data sets. Full article

High-quality observational datasets are essential for climate research and models, but validating and filtering decades of meteorological measurements is an enormous task. Advances in machine learning provide opportunities to expedite and improve quality control while offering insight into non-linear interactions between the meteorological variables. The Cabauw Experimental Site for Atmospheric Research in the Netherlands, known for its 213 m observation mast, has provided in situ observations for over 50 years. Despite high-quality instrumentation, measurement errors or non-representative data are inevitable. We explore machine-learning-assisted quality control, focusing on wind vane stalling at 10 m height. Wind vane stalling is treated as a binary classification problem as we evaluate five supervised methods (Logistic Regression, K-Nearest Neighbour, Random Forest, Gaussian Naive Bayes, Support Vector Machine) and one semi-supervised method (One-Class Support Vector Machine). Our analysis determines that wind vane stalling occurred 4.54% of the time annually over 20 years, often during stably stratified nocturnal conditions. The K-Nearest Neighbour and Random Forest methods performed the best, identifying stalling with approximately 75% accuracy, while others were more affected by data imbalance (more non-stalling than stalling data points). The semi-supervised method, avoiding the effects of the inherent data imbalance, also yielded promising results towards advancing data quality assurance. Full article

Flood susceptibility provides scientific support for flood prevention planning and infrastructure development by identifying and assessing flood-prone areas. The uncertainty posed by non-flood sample datasets remains a key challenge in flood susceptibility mapping. Therefore, this study proposes a novel sampling method for non-flood points. A flood susceptibility model is constructed using a machine learning algorithm to examine the uncertainty in flood susceptibility due to non-flood point selection. The influencing factors of flood susceptibility are analyzed through interpretable models. Compared to non-flood datasets generated by random sampling with the buffer method, the non-flood dataset constructed using the spatial range identified by the frequency ratio model and sampling method of one-class support vector machine achieves higher accuracy. This significantly improves the simulation accuracy of the flood susceptibility model, with an accuracy increase of 24% in the ENSEMBLE model. (2) In constructing the flood susceptibility model using the optimal non-flood dataset, the ENSEMBLE learning algorithm demonstrates higher accuracy than other machine learning methods, with an AUC of 0.95. (3) The northern and southeastern regions of the Zijiang River Basin have extremely high flood susceptibility. Elevation and drainage density are identified as key factors causing high flood susceptibility in these areas, whereas the southwestern region exhibits low flood susceptibility due to higher elevation. (4) Elevation, slope, and drainage density are the three most important factors affecting flood susceptibility. Lower values of elevation and slope and higher drainage density correlate with higher flood susceptibility. This study offers a new approach to reducing uncertainty in flood susceptibility and provides technical support for flood prevention and disaster mitigation in the basin. Full article

Predictive maintenance in manufacturing industry applications is a challenging research field. Packaging machines are widely used in a large number of logistic companies’ warehouses and must be working uninterruptedly. Traditionally, preventive maintenance strategies have been carried out to improve the performance of these machines. However, these kinds of policies do not take into account the information provided by the sensors implemented in the machines. This paper presents an expert system for the automatic estimation of work orders to implement predictive maintenance policies for packaging machines. The central innovation lies in a two-stage process: a classifier generates a binary decision on whether a machine requires maintenance, and an unsupervised anomaly detection module subsequently audits the classifier’s probabilistic output to refine and interpret its predictions. By leveraging the classifier to condense sensor data and applying anomaly detection to its output, the system optimizes the decision reliability. Three anomaly detection methods were evaluated: One-Class Support Vector Machine (OCSVM), Minimum Covariance Determinant (MCD), and a majority (hard) voting ensemble of the two. All anomaly detection methods improved the baseline classifier’s performance, with the majority voting ensemble achieving the highest F1 score. Full article

In the face of growing Internet of Things (IoT) security challenges, traditional Intrusion Detection Systems (IDSs) fall short due to IoT devices’ unique characteristics and constraints. This paper presents an effective, lightweight detection model that strengthens IoT security by addressing the high dimensionality of IoT data. This model merges an asymmetric stacked autoencoder with a Deep Neural Network (DNN), applying one-class learning. It achieves a high detection rate with minimal false positives in a short time. Compared with state-of-the-art approaches based on the BoT-IoT dataset, it shows a higher detection rate of up to 96.27% in 0.27 s. Also, the model achieves an accuracy of 99.99%, precision of 99.21%, and f1 score of 97.69%. These results demonstrate the effectiveness and significance of the proposed model, confirming its potential for reliable deployment in real IoT security problems. Full article

Detecting malware has become extremely important with the increasing exposure of computational systems and mobile devices to online services. However, the rapidly evolving nature of malicious software makes this task particularly challenging. Despite the significant number of machine learning works for malware detection proposed in the last few years, limited interest has been devoted to continual learning approaches, which could allow models to showcase effective performance in challenging and dynamic scenarios while being computationally efficient. Moreover, most of the research works proposed thus far adopt a fully supervised setting, which relies on fully labelled data and appears to be impractical in a rapidly evolving malware landscape. In this paper, we address malware detection from a continual semi-supervised one-class learning perspective, which only requires normal/benign data and empowers models with a greater degree of flexibility, allowing them to detect multiple malware types with different morphology. Specifically, we assess the effectiveness of two replay strategies on anomaly detection models and analyze their performance in continual learning scenarios with three popular malware detection datasets (CIC-AndMal2017, CIC-MalMem-2022, and CIC-Evasive-PDFMal2022). Our evaluation shows that replay-based strategies can achieve competitive performance in terms of continual ROC-AUC with respect to the considered baselines and bring new perspectives and insights on this topic. Full article

Multi-class anomaly detection is more efficient and less resource-consuming in industrial anomaly detection scenes that involve multiple categories or exhibit large intra-class diversity. However, most industrial image anomaly detection methods are developed for one-class anomaly detection, which typically suffer significant performance drops in multi-class scenarios. Research specifically targeting multi-class anomaly detection remains relatively limited. In this work, we propose a powerful unified normalizing flow for multi-class anomaly detection, which we call UniFlow. A multi-cognitive visual adapter (Mona) is employed in our method as the feature adaptation layer to adapt image features for both the multi-class anomaly detection task and the normalizing flow model, facilitating the learning of general knowledge of normal images across multiple categories. We adopt multi-cognitive convolutional networks with high capacity to construct the coupling layers within the normalizing flow model for more effective multi-class distribution modeling. In addition, we employ a multi-scale feature fusion module to aggregate features from various levels, thereby obtaining fused features with enhanced expressive capabilities. UniFlow achieves a class-average image-level AUROC of 99.1% and a class-average pixel-level AUROC of 98.0% on MVTec AD, outperforming the SOTA multi-class anomaly detection methods. Extensive experiments on three benchmark datasets, MVTec AD, VisA, and BTAD, demonstrate the efficacy and superiority of our unified normalizing flow in multi-class anomaly detection. Full article

The usage of anomaly detection is of critical importance to numerous domains, including structural health monitoring (SHM). In this study, we examine an online setting for damage detection in the Z24 bridge. We evaluate and compare the performance of the elliptic envelope, incremental one-class support vector classification, local outlier factor, half-space trees, and entropy-guided envelopes. Our findings demonstrate that XGBoost exhibits enhanced performance in identifying a limited set of significant features. Additionally, we present a novel approach to manage drift through the application of entropy measures to structural state instances. The study is the first to assess the applicability of one-class classification for anomaly detection on the short-term structural health data of the Z24 bridge. Full article

The safe operation of aero-engines is crucial for ensuring flight safety, and effective fault detection methods are fundamental to achieving this objective. In this paper, we propose a novel approach that integrates an auto-encoder with long short-term memory (LSTM) networks and a self-attention mechanism for the anomaly detection of aero-engine time-series data. The dataset utilized in this study was simulated from real data and injected with fault information. A fault detection model is developed utilizing normal data samples for training and faulty data samples for testing. The LSTM auto-encoder processes the time-series data through an encoder–decoder architecture, extracting latent representations and reconstructing the original inputs. Furthermore, the self-attention mechanism captures long-range dependencies and significant features within the sequences, thereby enhancing the detection accuracy of the model. Comparative analyses with the traditional LSTM auto-encoder, as well as one-class support vector machines (OC-SVM) and isolation forests (IF), reveal that the experimental results substantiate the feasibility and effectiveness of the proposed method, highlighting its potential value in engineering applications. Full article

Idlers are essential to conveyor systems, as well as supporting and guiding belts to ensure production efficiency. Proper idler maintenance prevents failures, reduces downtime, cuts costs, and improves reliability. Most studies on idler fault detection rely on supervised methods, which depend on large labelled datasets for training. However, acquiring such labelled data is often challenging in industrial environments due to the rarity of faults and the labour-intensive nature of the labelling process. To address this, we propose the chroma-augmented semi-supervised anomaly detection (CASSAD) method, designed to perform effectively with limited labelled data. At the core of CASSAD is the one-class SVM (OC-SVM), a model specifically developed for anomaly detection in cases where labelled anomalies are scarce. We also compare CASSAD’s performance with other common models like the local outlier factor (LOF) and isolation forest (iForest), evaluating each with the area under the curve (AUC) to assess their ability to distinguish between normal and anomalous data. CASSAD introduces chroma features, such as chroma energy normalised statistics (CENS), the constant-Q transform (CQT), and the chroma short-time Fourier transform (STFT), enhanced through filtering to capture rich harmonic information from idler sounds. To reduce feature complexity, we utilize the mean and standard deviation (std) across chroma features. The dataset is further augmented using additive white Gaussian noise (AWGN). Testing on an industrial dataset of idler sounds, CASSAD achieved an AUC of 96% and an accuracy of 91%, surpassing a baseline autoencoder and other traditional models. These results demonstrate the model’s robustness in detecting anomalies with minimal dependence on labelled data, offering a practical solution for industries with limited labelled datasets. Full article

A harmonic reducer is an important component of industrial robots. In practical applications, it is difficult to obtain enough anomaly data from error cases for the supervised training of models. Whether the information contained in regular features is sensitive to anomaly detection is unknown. In this paper, we propose an anomaly detection frame for a harmonic reducer with only healthy data. We considered an auto-encoder trained using only healthy features, such as feature mapping, in which the difference between the output and the input constitutes a new high-dimensional feature space that retained information relevant only to anomalies. Compared to the original feature space, this space was more sensitive to abnormal data. The mapped features were then fed into the OCSVM to preserve the feature details of the abnormal information. The effectiveness of this method was validated by multiple sets of data collecting from harmonic reducers. Three different residual calculations and four different AE models were used, showing that the method outperforms an AE or an OCSVM alone. It is also verified that the method outperforms other typical anomaly detection methods. Full article

The analysis of fast fluorescence kinetics, specifically through the JIP test, is a valuable tool for identifying and characterizing plant stress. However, interpreting OJIP data requires a comprehensive understanding of their underlying theory. This study proposes a Machine Learning-based approach using a One-Class Support Vector Machine anomaly detection model to effectively categorize OJIP measurements into “normal”, representing healthy plants, and “anomalies”. This approach was validated using a previously published dataset. A subgroup of the identified “anomalies” was clearly linked to stress-induced reductions in photosynthesis. Furthermore, the percentage of these “anomalies” showed a meaningful correlation with both the progression and severity of stress. The results highlight the still largely unexploited potential of Machine Learning in OJIP analysis. Full article