Anomaly Detection

Background: Anomaly detection is a critical aspect of data analysis, demanding rigorous methodologies to ensure accuracy and reliability. In this paper, we present an anomaly detection algorithm specifically engineered for detecting anomalies within rotated elliptical clusters. Objectives: Our study aims to bridge the gap between theoretical insights and practical applications by leveraging precise geometric computations and robust statistical frameworks to achieve unparalleled levels of accuracy and reliability. Methods: Central to our algorithm is a heuristic that establishes definitive criteria for identifying anomalies outside rotated elliptical clusters, supported by rigorous mathematical analysis. Results: Through meticulous mathematical modeling, our algorithm consistently outperforms existing methods by 20-30%, showcasing its superiority in practical applications. Rigorous testing and validation across diverse datasets demonstrate an unprecedented level of performance, with recall, precision, and F1 score approaching 1.0. Conclusions: By providing a comprehensive solution tailored to the nuances of anomaly detection within elliptical clusters, our research represents a significant advancement in anomaly detection methodologies. Our algorithm's near-perfect accuracy rate, backed by concrete mathematical proofs and empirical validation, establishes a new benchmark and paves the way for enhanced data analysis techniques. We anticipate widespread adoption in various domains, where precise anomaly detection is paramount.

Conventional dimensionality reduction methods like Multidimensional Scaling (MDS) are sensitive to the presence of orthogonal outliers, leading to significant defects in the embedding. We introduce a robust MDS method, called DeCOr-MDS (Detection and Correction of Orthogonal outliers using MDS), based on the geometry and statistics of simplices formed by data points, that allows to detect orthogonal outliers and subsequently reduce dimensionality. We validate our methods using synthetic datasets, and further show how it can be applied to a variety of large real biological datasets, including cancer image cell data and human microbiome project data.

We present a tool for Interactive Visual Exploration of Latent Space (IVELS) for model selection. Evaluating generative models of discrete sequences from a continuous latent space is a challenging problem, since their optimization involves multiple competing objective terms. We introduce a model-selection pipeline to compare and filter models throughout consecutive stages of more complex and expensive metrics. We present the pipeline in an interactive visual tool to enable the exploration of the metrics, analysis of the learned latent space, and selection of the best model for a given task. We focus specifically on the variational auto-encoder family in a case study of modeling peptide sequences, which are short sequences of amino acids. This task is especially interesting due to the presence of multiple attributes we want to model. We demonstrate how an interactive visual comparison can assist in evaluating how well an unsupervised auto-encoder meaningfully captures the attributes ...

Dear Colleagues,

The integration of machine learning (ML) with image and signal processing has revolutionized various fields by enhancing researchers’ ability to analyse and interpret complex data. ML algorithms can extract meaningful features from signals and images, improve classification accuracy, and enable advanced functionalities such as real-time processing and automated decision-making. This synergy between ML and image/signal processing has broad applications, including medical diagnostics, remote sensing, security, and multimedia.

This Special Issue highlights innovative research and addresses ongoing challenges in this field, with a particular focus on signal and image processing, signal reconstruction, image quantization, and advanced medical imaging applications. It serves as a platform for researchers to share their cutting-edge research, insights, and experiences, fostering collaboration and advancing the field.

This Special Issue’s scope includes a wide range of topics related to the machine learning applications in image and signal processing. Topics of interest include, but are not limited to, the following:

    Pre-trained models for imaging diagnosis;
    Image segmentation, quantization, classification and dimensionality reduction;
    Pattern recognition;
    Molecular imaging and nuclear medicine;
    X-ray, CT-scan, MRI, NMR, FMRI, and ultrasonography analysis;
    Machine learning applications in OCT, OCTA, and other ophthalmology imaging
    Biometric recognition systems;
    Integration of machine learning in imaging devices, imaging sequences, and imaging-guided intervention;
    Anomaly detection, signal denoising, and signal classification;
    Signal filtering and reconstruction.

Dr. Siamak Pedrammehr
Guest Editor

How to cite this article Wijaya WM, Nakamura Y. 2023. Loitering behavior detection by spatiotemporal characteristics quantification based on the dynamic features of Automatic Identification System (AIS) messages. PeerJ Comput. Sci. 9:e1572 http://doi.org/10.7717/peerjcs.1572 km2. The results prove the efficacy of the proposed method. It remarkably outperforms the existing approach with 97% accuracy and 92% F-score. The experiment produces a ranked list of loitering vessels and an intuitive visualization in the relevant geographic area. In the realworld scenario, they are practical means to support further examination by human operators.

In this paper, we aimed to help bridge the gap between human fluid intelligence-the ability to solve novel tasks without prior training-and the performance of deep neural networks, which typically require extensive prior training. An essential cognitive component for solving intelligence tests, which in humans are used to measure fluid intelligence, is the ability to identify regularities in sequences. This motivated us to construct a benchmark task, which we term sequence consistency evaluation (SCE), whose solution requires the ability to identify regularities in sequences. Given the proven capabilities of deep networks, their ability to solve such tasks after extensive training is expected. Surprisingly, however, we show that naive (randomly initialized) deep learning models that are trained on a single SCE with a single optimization step can still solve non-trivial versions of the task relatively well. We extend our findings to solve, without any prior training, realworld anomaly detection tasks in the visual and auditory modalities. These results demonstrate the fluid-intelligent computational capabilities of deep networks. We discuss the implications of our work for constructing fluid-intelligent machines.

Artificial intelligence and machine learning are widely used and have significant benefits in various areas of knowledge, including healthcare. However, there is an important barrier to disseminating machine learning among healthcare professionals, which consists mainly of their unfamiliarity with programming and computing concepts. The development strategy called 'low-code,' when applied to software development, includes frameworks and tools that, in short, make application development more accessible to professional communities. The low-code strategy simplifies the software development process. This strategy is relevant for smart cities that aim to develop approaches that can make cities more efficient, humane, and sustainable and contribute to achieving the sustainable development goals (SDGs) of the United Nations (UN). This article positions the low-code strategy through the PyCaret framework as an element of innovation and contribution to the development of health systems using machine learning in smart cities. The paper presents the low-code strategy through a case study that evaluates the incidence and correlation of occupational activities with the occurrence of cancer in Brazil using an anomaly detection algorithm. The article's contributions include positioning the low-code strategy as an element of innovation in smart cities and presenting a case study that can be a reference for developing applications with machine learning in the healthcare area. The case study presented, in turn, presents a differentiated approach to detecting cancer using an anomaly detection algorithm and reiterates correlations between types of cancer and occupational activities.

In an era where organizations increasingly rely on intricate software applications, cloud services, and interconnected networks, the significance of analyses cannot be overstated. These tools serve as vigilant custodians of digital footprints, meticulously dissecting the voluminous records encapsulated in log files to extract valuable insights and detect anomalies. As such, log analyses emerge as linchpins in deciphering the meaning behind recorded events, enabling organizations to obtain a more comprehensive understanding of their digital infrastructures and enhance their security posture. This comprehensive exploration aims to untangle the core attributes of log analyses, bringing clarity to their parsing capabilities, the art of information extraction, and the nuanced algorithms that facilitate the conversion of raw logs into actionable insights. Furthermore, against the backdrop of a dynamically evolving cyber threat landscape, the role of log analyses extends beyond conventional diagnostics. These tools have become instrumental in the proactively orchestrated defense against cyber adversaries, empowering organizations to detect and mitigate threats in real-time. Through an in-depth analysis of log analyses and their evolving functionalities, this paper seeks to provide a comprehensive understanding of their Integral role in modern cybersecurity and system management. By elucidating the significance and impact of log analyses, organizations can leverage these tools to fortify their defenses, mitigate risks, and make informed, data-driven decisions in an increasingly complex digital environment.

Production monitoring in real-time is a very important problem in smart manufacturing. It helps enterprises to timely detect abnormalities in the production process and then guarantee the product quality and reduce waste. In this paper, we develop a novel method to monitor the real-time production based on the Convolution Neural Network and the Support Vector Data Description algorithm. The numerical result shows that our proposed method leads to high efficient on the testing data.

The rapid development of the Internet and the wide spread of its applications has affected many aspects of our life. However, this development also makes the cyberspace more vulnerable to various attacks. Thus, detecting and preventing these attacks are crucial for the next development of the Internet and its services. Recently, machine learning methods have been widely adopted in detecting network attacks. Among many machine learning methods, AutoEncoders (AEs) are known as the state-of-the-art techniques for network anomaly detection. Although, AEs have been successfully applied to detect many types of attacks, it is often unable to detect some difficult attacks that attempt to mimic the normal network traffic. In order to handle this issue, we propose a new model based on AutoEncoder called Double-Shrink AutoEncoder (DSAE). DSAE put more shrinkage on the normal data in the middle hidden layer. This helps to pull out some anomalies that are very similar to normal data. DSAE are ev...

In the rapidly evolving field of software development, the need for efficient and effective testing methods is paramount. Traditional test automation has been a cornerstone in achieving reliable software delivery, but it is not without its limitations. Enter machine learning (ML)-a technology poised to revolutionize test automation by making it smarter, faster, and more adaptive. This paper explores the integration of ML in test automation, presenting a framework for its implementation and evaluating its impact through experimental results.

This paper was written to test the EMH(Efficient Market Hypothesis),created by Eugene Fama, on the current stock market. EMH is a famous hypothesis that states the stock price reflects all available information in the market, thus not leaving any room for investors to make money. However, in reality, there are plenty of people that are benefiting off of the stock market unlike the efficient market that Fama has theorized. To test our hypothesis where the current stock market is in the weaker side of the EMH strength wise, supporting the theorized prediction of the strength of the market, we developed a ARIMA(Autoregression Integrated Moving Average) Model with Information-based predictor added to it which its residual was compared to the original ARIMA models residual calculated from the same data. The results shows that the residual of the newly developed model was statistically smaller, meaning that good news and bad news can help traders create profit just like in theory, making the stock market a weaker form market of the Efficient Market Hypothesis.

In the budding cyber security threats, Botnet represent the extremely critical threats being encountered by several organizations recently. Botnets are used to carry out many cybercrimes reported in the recent news. Although significantly many explorations have been undertaken on Botnet evaluation and discovery, a number of faces stayed unaddressed, for instance the competence to model detectors that deal with innovative types of Botnets. In this article, we are proposing a new Botnet detection system named BOTRANSACK using ML-centered classification skills to classify the command and control (C & C) traffic of IRC-based Botnets. Subsequently the identification process, the scheme will engender sound alarm and send text message to the admin of the system who can activate utmost suitable security activities, like hindering the corresponding IP address, placing them beneath a deeper scrutiny or acting over same apprehensive network segments.

Anomaly detection is very important for system monitoring and security since successful execution of these engineering tasks depends on access to validated data. The localization of the variable causing the fault is very essential. Indeed, the localization of the fault is defined as the ability to determine the source of the fault on a system. Generally, the identification of faults is linked to the detection procedure implemented. Therefore, it is very important to choose the adequate fault detection model to locate fault. For nonlinear uncertain systems, the most performed fault detection method is reduced rank interval kernel principal component analysis (RRIKPCA), which enhances the computational skill by downgrading the kernel matrix dimension. We have proposed in this article a new fault localization technique for uncertain systems, named partial RRIKPCA, which combines the benefits of the RRIKPCA technique and the principle of partial localization. The principal of this metho...

The use of surveillance cameras is a common solution that addresses the need to provide security and manage urban traffic that arises due to the increasing population in cities. As the number of surveillance cameras rises, video streams that create big data are recorded. The analysis of video streams collected from those traffic surveillance cameras and the automatic detection of unusual, suspicious events, as well as a range of harmful activities, have become crucial because it is impossible to observe, analyze, and comprehend the contents of these movies using human labor. Recent studies have shown that deep learning (DL)-based artificial intelligence (AI) techniques, particularly machine learning (ML) systems, are used in video anomaly detection (VAD) studies. In this study, an efficient frame-level VAD method is proposed based on transfer learning (TL) and fine-tuning (FT) approach, and anomalies were detected using 20 popular convolutional neural network (CNN)-based DL models where variants of VGG, Xception, MobileNet, Inception, EfficientNet, ResNet, DenseNet, NASNet, and ConvNeXt base models were trained via the TL and FT approaches. The proposed approach was tested using the CUHK Avenue, UCSD Ped1, and UCSD Ped2 datasets, and the performances of the models were measured via area under curve (AUC), accuracy, precision, recall, and F1-score metrics. The highest AUC scores measured were 100%, 100%, and 98.41% for the UCSD Ped1, UCSD Ped2, and CUHK Avenue datasets, respectively. Compared to existing techniques in the literature, experimental results show that the suggested method offers state-of-the-art VAD performance.

Purpose Surgical workflow recognition is a crucial and challenging problem when building a computer-assisted surgery system. Current techniques focus on utilizing a convolutional neural network and a recurrent neural network (CNN–RNN) to solve the surgical workflow recognition problem. In this paper, we attempt to use a deep 3DCNN to solve this problem. Methods In order to tackle the surgical workflow recognition problem and the imbalanced data problem, we implement a 3DCNN workflow referred to as I3D-FL-PKF. We utilize focal loss (FL) to train a 3DCNN architecture known as Inflated 3D ConvNet (I3D) for surgical workflow recognition. We use prior knowledge filtering (PKF) to filter the recognition results. Results We evaluate our proposed workflow on a large sleeve gastrectomy surgical video dataset. We show that focal loss can help to address the imbalanced data problem. We show that our PKF can be used to generate smoothed prediction results and improve the overall accuracy. We sh...