Explore the vast range of titles available.

While many structural damage detection methods have been developed in recent decades, few data‐driven methods in unsupervised learning mode have been developed to solve the practical difficulties in data acquisition for civil infrastructures in different scenarios. To address such a challenge, this article proposes a number of improved unsupervised novelty detection methods and conducts extensive comparative studies on a laboratory scale steel bridge to examine their performances of damage detection. The key concept behind unsupervised novelty detection in this article is that only normal data from undamaged/baseline structural scenarios are required to train statistical models with these methods. Then, these trained models are used to identify abnormal testing data from damaged scenarios. To detect structural damage in the form of loosening bolts in the steel bridge, four machine‐learning methods (i.e., K‐nearest neighbors method, Gaussian mixture models, one‐class support vector machines, density peaks‐based fast clustering method) and one deep learning method using a deep auto‐encoder are selected. Meanwhile, some modifications and improvements are made to enable these methods to detect structural damage in unsupervised novelty detection mode. In their comparative studies, the advantages and disadvantages of these methods are analyzed based on their results of structural damage detection. Recently, deep learning‐based damage detection is a very hot topic. This article conducted extensive comparative studies using state‐of‐the‐art methods of deep learning‐based damage detection methods to figure out the pros and cons of each method.

The unsupervised detection and localization of anomalies in natural images is an intriguing and challenging problem. Anomalies manifest themselves in very different ways and an ideal benchmark dataset for this task should contain representative examples for all of them. We find that existing datasets are biased towards local structural anomalies such as scratches, dents, or contaminations. In particular, they lack anomalies in the form of violations of logical constraints, e.g., permissible objects occurring in invalid locations. We contribute a new dataset based on industrial inspection scenarios that evenly covers both types of anomalies. We provide pixel-precise ground truth data for each anomalous region and define a generalized evaluation metric that addresses localization ambiguities that can arise for logical anomalies. Furthermore, we propose a novel algorithm that improves over the state of the art in the joint detection of structural and logical anomalies. It consists of a local and a global network branch. The first one inspects confined regions independent of their spatial locations in the input image and is primarily responsible for the detection of entirely new local structures. The second one learns a globally consistent representation of the training data through a bottleneck that enables the detection of violations of long-range dependencies, a key characteristic of many logical anomalies. We perform extensive evaluations on our new dataset to corroborate our claims.

The detection of anomalous structures in natural image data is of utmost importance for numerous tasks in the field of computer vision. The development of methods for unsupervised anomaly detection requires data on which to train and evaluate new approaches and ideas. We introduce the MVTec anomaly detection dataset containing 5354 high-resolution color images of different object and texture categories. It contains normal, i.e., defect-free images intended for training and images with anomalies intended for testing. The anomalies manifest themselves in the form of over 70 different types of defects such as scratches, dents, contaminations, and various structural changes. In addition, we provide pixel-precise ground truth annotations for all anomalies. We conduct a thorough evaluation of current state-of-the-art unsupervised anomaly detection methods based on deep architectures such as convolutional autoencoders, generative adversarial networks, and feature descriptors using pretrained convolutional neural networks, as well as classical computer vision methods. We highlight the advantages and disadvantages of multiple performance metrics as well as threshold estimation techniques. This benchmark indicates that methods that leverage descriptors of pretrained networks outperform all other approaches and deep-learning-based generative models show considerable room for improvement.

Structural damage detection using unsupervised learning methods has been a trending topic in the structural health monitoring (SHM) research community during the past decades. In the context of SHM, unsupervised learning methods rely only on data acquired from intact structures for training the statistical models. Consequently, they are often seen as more practical than their supervised counterpart in implementing an early-warning damage detection system in civil structures. In this article, we review publications on data-driven structural health monitoring from the last decade that relies on unsupervised learning methods with a focus on real-world application and practicality. Novelty detection using vibration data is by far the most common approach for unsupervised learning SHM and is, therefore, given more attention in this article. Following a brief introduction, we present the state-of-the-art studies in unsupervised-learning SHM, categorized by the types of used machine-learning methods. We then examine the benchmarks that are commonly used to validate unsupervised-learning SHM methods. We also discuss the main challenges and limitations in the existing literature that make it difficult to translate SHM methods from research to practical applications. Accordingly, we outline the current knowledge gaps and provide recommendations for future directions to assist researchers in developing more reliable SHM methods.

In this work, we propose a one-class self-supervised method for anomaly segmentation in images that benefits from both a modern machine learning approach and a more classic statistical detection theory. The method consists of four phases. First, features are extracted using a multi-scale image transformer architecture. Then, these features are fed into a U-shaped normalizing flow (NF) that lays the theoretical foundations for the subsequent phases. The third phase computes a pixel-level anomaly map from the NF embedding, and the last phase performs a segmentation based on the a contrario framework. This multiple hypothesis testing strategy permits the derivation of robust unsupervised detection thresholds, which are crucial in real-world applications where an operational point is needed. The segmentation results are evaluated using the mean intersection over union metric, and for assessing the generated anomaly maps we report the area under the receiver operating characteristic curve ( AUROC ), as well as the area under the per-region-overlap curve ( AUPRO ). Extensive experimentation in various datasets shows that the proposed approach produces state-of-the-art results for all metrics and all datasets, ranking first in most MVTec-AD categories, with a mean pixel-level AUROC of 98.74%. Code and trained models are available at https://github.com/mtailanian/uflow .

This study investigates the impact of data distribution and bootstrap resampling on the anomaly detection performance of the Isolation Forest (iForest) algorithm in statistical process control. Although iForest has received attention for its multivariate and ensemble-based nature, its performance under non-normal data distributions and varying bootstrap settings remains underexplored. To address this gap, a comprehensive simulation was performed across 18 scenarios involving log-normal, gamma, and t-distributions with different mean shift levels and bootstrap configurations. The results show that iForest substantially outperforms the conventional Hotelling’s T2 control chart, especially in non-Gaussian settings and under small-to-medium process shifts. Enabling bootstrap resampling led to marginal improvements across classification metrics, including accuracy, precision, recall, F1-score, and average run length (ARL)1. However, a key limitation of iForest was its reduced sensitivity to subtle process changes, such as a 1σ mean shift, highlighting an area for future enhancement.

Novelty detection is one of the classic problems in machine learning that has applications across several domains. This paper proposes a novel autoencoder based on Deep Gaussian Processes for novelty detection tasks. Learning the proposed model is made tractable and scalable through the use of random feature approximations and stochastic variational inference. The result is a flexible model that is easy to implement and train, and can be applied to general novelty detection tasks, including large-scale problems and data with mixed-type features. The experiments indicate that the proposed model achieves competitive results with state-of-the-art novelty detection methods.

Novelty Detection is a task of recognition of abnormal data points within a given system. Recently, this task has been performed using Deep Learning Autoencoders, but they face several drawbacks which include the problem of identity mapping, adversarial perturbations and optimization algorithms. In this paper, we have proposed a novel approach LPRNet, a Denoising Autoencoder which uses algorithms such as Least Trimmed Square, Projected Gradient Descent and Robust Principal Component Analysis, to solve the above-mentioned problems. LRPNet is then trained and tested on NSL-KDD dataset, and experiments have been performed using Accuracy as performance metric for comparing the existing models with the proposed model. The results show that LRPNet has the maximum accuracy of 95.9% and performed better than all the previous state-of-the-art algorithms.

In statistical learning theory, numerous works established non-asymptotic bounds assessing the generalization capacity of empirical risk minimizers under a large variety of complexity assumptions for the class of decision rules over which optimization is performed, by means of sharp control of uniform deviation of i.i.d. averages from their expectation, while fully ignoring the possible dependence across training data in general. It is the purpose of this paper to show that similar results can be obtained when statistical learning is based on a data sequence drawn from a (Harris positive) Markov chain X , through the running example of estimation of minimum volume sets (MV-sets) related to X ’s stationary distribution, an unsupervised statistical learning approach to anomaly/novelty detection. Based on novel maximal deviation inequalities we establish, using the regenerative method , learning rate bounds that depend not only on the complexity of the class of candidate sets but also on the ergodicity rate of the chain X , expressed in terms of tail conditions for the length of the regenerative cycles. In particular, this approach fully tailored to Markovian data permits to interpret the rate bound results obtained in frequentist terms, in contrast to alternative coupling techniques based on mixing conditions: the larger the expected number of cycles over a trajectory of finite length, the more accurate the MV-set estimates. Beyond the theoretical analysis, this phenomenon is supported by illustrative numerical experiments.

Temporal segmentation of videos into meaningful image sequences containing some particular activities is an interesting problem in computer vision. We present a novel algorithm to achieve this semantic video segmentation. The segmentation task is accomplished through event detection in a frame-by-frame processing setup. We propose using one-class classification (OCC) techniques to detect events that indicate a new segment, since they have been proved to be successful in object classification and they allow for unsupervised event detection in a natural way. Various OCC schemes have been tested and compared, and additionally, an approach based on the temporal self-similarity maps (TSSMs) is also presented. The testing was done on a challenging publicly available thermal video dataset. The results are promising and show the suitability of our approaches for the task of temporal video segmentation.