One-class Support Vector Machine Research Articles

A Health Status Assessment Method of Fluid Loop System Based on Hierarchical Multi-Information Fusion

As a kind of thermal control device, fluid loop systems must operate with the demands of high safety, high reliability, and long life. In order to accurately assess the health status of fluid loop systems, a hierarchical multi-information fusion (HMIF) method is proposed in this paper. Considering that fluid loop systems generally have distinct structural hierarchies, the health evaluation process in this method is divided into three levels, which are the indicator level, the component level, and the system level. In the evaluation process, the health indices are, respectively, constructed to quantify the health status at the three levels. At the indicator level, One-Class support vector machine algorithm is used to obtain the distribution space of each state monitoring indicator under a normal state. The indicator-level health indices are evaluated by calculating the ratio of the data located in the distribution space to the overall data. At the component level, a fuzzy theory is used to calculate the health indices of the component level. Health indices at the indicator level are first converted to membership degree by membership degree function. Then, the evaluation fusion strategy is used to deduce the membership degree of the component level. The health indices at the component level are obtained from the mapping relationship between the membership degree and the health index. At the system level, an adaptive weight adjustment strategy is proposed to fuze all component-level health indices. Taking a practical fluid loop system as an example, health indices at the three levels evaluated by the HMIF method are compared with the actual status. The results indicate that the proposed method can correctly judge the health state of the system and provide a reference for the maintenance and fault diagnosis of fluid loop systems.

Machine Learning Approaches to Advanced Outlier Detection in Psychological Datasets

The core aim of this study is to determine the most effective outlier detection methodologies for multivariate psychological datasets, particularly those derived from Omani students. Due to their complex nature, such datasets demand robust analytical methods. To this end, we employed three sophisticated algorithms: local outlier factor (LOF), one-class support vector machine (OCSVM), and isolation forest (IF). Our initial findings showed 155 outliers by both LOF and IF and 147 by OCSVM. A deeper analysis revealed that LOF detected 55 unique outliers based on differences in local density, OCSVM isolated 44 unique outliers utilizing its transformed feature space, and IF identified 76 unique outliers leveraging its tree-based mechanics. Despite these varying results, all methods had a consensus for just 44 outliers. Employing ensemble techniques, both averaging and voting methods identified 155 outliers, whereas the weighted method highlighted 151, with a consensus of 150 outliers across the board. In conclusion, while individual algorithms provide distinct perspectives, ensemble techniques enhance the accuracy and consistency of outlier detection. This underscores the necessity of using multiple algorithms with ensemble techniques in analyzing psychological datasets, facilitating a richer comprehension of inherent data structures.

OS-FCM: A semi-supervised clustering approach to investigating consumers’ usage patterns of contactless shopping-delivery (S-D) channel

Investigating the factors that contribute to consumers’ usage patterns of contactless shopping-delivery (S-D) channel is important for e-commerce retailers and logistics operators. However, in the data as collected by a questionnaire survey, distinct usage patterns are not always easily observed due to the existence of outliers, which places a challenge for the investigation. Hence, this study develops a labelling approach, named OS-FCM, to determine the usage patterns. As a semi-supervised clustering approach, OS-FCM integrates One-Class Support Vector Machine and Fuzzy C-Means, which enables it to make a trade-off between subjective judgement and statistical evidence when identifying patterns and to effectively recognize the patterns of outliers. The dataset with the usage patterns is used to train Random Forest, upon which Shapley Additive exPlanations (SHAP) approach is applied to measure the contributions of each factor. The data as collected by a questionnaire survey in Singapore are used for a case study. By using OS-FCM, the participants are clustered into five patterns with each pattern involving three variables, namely, current usage frequency, current behavioral habit, and future continuance intention. The contributions of each potential factor to the patterns are measured by SHAP values. In the case study, two regular patterns and three outlier patterns are identified. It is found that consumers’ past behavior stands out to be the most important factor to the pattern formation from an aggregated perspective. Factors such as fear appeal and behavior maintenance/change motivations exert differentiated impacts on the formation of the five patterns. Besides, as compared to commonly used clustering methods, OS-FCM has better clustering performance and the patterns determined by OS-FCM are more reasonable and rational, which provides a solid basis for better understanding the usage patterns.

Deep Anomaly Detection with Attention (DADA): A Novel Approach for Identifying Multipath Interference in Radar Signals

Multipath interference in radar signals caused by sea, ground, and other environments poses significant challenges to the target detection, tracking, and classification capabilities of radar systems. Existing methods for radar signal identification require labeled samples and focus mainly on the classification of normal signals. However, in practice, anomalous samples (multipath interference signals) may be scarce and highly imbalanced (i.e., mostly normal samples). To address this problem, we propose a deep anomaly detection with attention (DADA) for semisupervised detection of multipath radar signals. The method transforms radar signals into time–frequency images and is trained exclusively on normal samples. The autoencoder architecture is extended with a feature extractor network to capture latent sample features. CBAM attention is introduced to improve feature extraction. By learning the distribution of normal samples in high-dimensional image space and low-dimensional feature space, a two-dimensional feature space representing normal samples is constructed. A one-class SVM then learns the boundary of normal samples for anomaly detection. Extensive experiments on radar signal datasets validate the effectiveness of the proposed approach.

Real-time Cyber-Physical Security Solution Leveraging an Integrated Learning-Based Approach

Cyber-Physical Systems (CPS) has emerged as a paradigm that connects cyber and physical worlds, which provides unprecedented opportunities to realize intelligent applications such as smart home, smart cities, and smart manufacturing. However, CPS faces a great number of information security challenges (e.g., attacks) due to the integration of CPS as well as the human behaviors and interactions. Therefore, accurate and real-time attack detection and identification are essential to ensure information security and reliability of CPS. In this paper, we propose a novel integrated learning method that accurately detects an attack of a CPS system and then identifies the attack type in real time. Specifically, we consider a One-Class Support Vector Machine (OCSVM) model that only relies on the data from the normal state for training to achieve a real-time and effective detection of a CPS system state (i.e., normal or under-attack). If the system is detected to be under-attack, we then develop a Pairwise Self-supervised Long Short-Term Memory (PSLSTM) approach to identify the attack type, which aims to accurately distinguish the known attack types and discover unknown new attacks. Lastly, experimental results show the proposed method achieves promising performances compared with conventional and state-of-the-art learning-based benchmarks.

One-Class SVM Probabilistic Outputs.

One-class support vector machine (SVM) is an extension of SVM to handle unlabeled data. As a mature technique for outlier detection, one-class SVM has been widely used in many applications. However, similar to standard two-class SVM, the design of one-class SVM does not give probabilistic outputs. For two-class SVM, some methods have been proposed to effectively obtain probabilistic outputs, but due to the difficulty of no-label information, less attention has been paid to one-class SVM. Our aim in this work is to propose some practically viable techniques to generate probabilistic outputs for one-class SVM. We investigate existing methods for two-class SVM and explain why they may not be suitable for one-class SVM. Due to the lack of label information, we think a feasible setting is to have probabilities mimic to the decision values of training data. Based on this principle, we propose several new methods. Detailed experiments on both artificial and real-world data demonstrate the effectiveness of the proposed methods.

Disruptive technology identification research based on one-class SVM: a case study in the field of gene editing

The identification of disruptive technologies is paramount for the expansion into novel markets and the creation of innovative value paradigms. Despite the importance of this task, current methodologies for identifying such technologies are predominantly qualitative, which are often hindered by a dearth of empirical data and a predisposition towards subjectivity. This paper proposes a paradigm shift by employing patent text data and utilizing Latent Dirichlet Allocation (LDA) for data preprocessing to extract and analyze technical themes from patent abstracts. The research innovatively posits the problem of identifying disruptive technologies as an anomaly detection issue and addresses it through the application of a one-class Support Vector Machine (SVM) algorithm. In the empirical component of the study, the field of gene editing is selected as a case study to validate the proposed method’s efficacy and practicality. The validation is conducted through a multifaceted analytical framework, which includes: an examination of the distribution and proportion of patent topics, a comprehensive market analysis, and a comparative analysis of the proposed model against existing methodologies. The results of this study not only substantiate the potential of the proposed approach in accurately detecting technologies with disruptive potential but also highlight its robustness across various analytical dimensions.

Application-layer Ddos Detection Based on a One-class Support Vector Machine

Application-layer Ddos Detection Based on a One-class Support Vector Machine

Hybrid computing framework security in dynamic offloading for IoT-enabled smart home system.

In the distributed computing era, cloud computing has completely changed organizational operations by facilitating simple access to resources. However, the rapid development of the IoT has led to collaborative computing, which raises scalability and security challenges. To fully realize the potential of the Internet of Things (IoT) in smart home technologies, there is still a need for strong data security solutions, which are essential in dynamic offloading in conjunction with edge, fog, and cloud computing. This research on smart home challenges covers in-depth examinations of data security, privacy, processing speed, storage capacity restrictions, and analytics inside networked IoT devices. We introduce the Trusted IoT Big Data Analytics (TIBDA) framework as a comprehensive solution to reshape smart living. Our primary focus is mitigating pervasive data security and privacy issues. TIBDA incorporates robust trust mechanisms, prioritizing data privacy and reliability for secure processing and user information confidentiality within the smart home environment. We achieve this by employing a hybrid cryptosystem that combines Elliptic Curve Cryptography (ECC), Post Quantum Cryptography (PQC), and Blockchain technology (BCT) to protect user privacy and confidentiality. Additionally, we comprehensively compared four prominent Artificial Intelligence anomaly detection algorithms (Isolation Forest, Local Outlier Factor, One-Class SVM, and Elliptic Envelope). We utilized machine learning classification algorithms (random forest, k-nearest neighbors, support vector machines, linear discriminant analysis, and quadratic discriminant analysis) for detecting malicious and non-malicious activities in smart home systems. Furthermore, the main part of the research is with the help of an artificial neural network (ANN) dynamic algorithm; the TIBDA framework designs a hybrid computing system that integrates edge, fog, and cloud architecture and efficiently supports numerous users while processing data from IoT devices in real-time. The analysis shows that TIBDA outperforms these systems significantly across various metrics. In terms of response time, TIBDA demonstrated a reduction of 10-20% compared to the other systems under varying user loads, device counts, and transaction volumes. Regarding security, TIBDA's AUC values were consistently higher by 5-15%, indicating superior protection against threats. Additionally, TIBDA exhibited the highest trustworthiness with an uptime percentage 10-12% greater than its competitors. TIBDA's Isolation Forest algorithm achieved an accuracy of 99.30%, and the random forest algorithm achieved an accuracy of 94.70%, outperforming other methods by 8-11%. Furthermore, our ANN-based offloading decision-making model achieved a validation accuracy of 99% and reduced loss to 0.11, demonstrating significant improvements in resource utilization and system performance.

Anomaly Detection in Binary Time Series Data: An unsupervised Machine Learning Approach for Condition Monitoring

A key element of smart manufacturing is condition monitoring and heath controlling of production machines. In today's rapidly evolving landscape of industrial machinery and equipment, optimizing the operation of production lines is critical to ensure high productivity and product quality. Timely detection and prevention of faults in the production process plays a crucial role in minimizing downtime, reducing costs, and ensuring optimal performance. The scientific challenge here is that the increasing number of sensors and actuators with digital input and output signals in the production machines creates different patterns, which are difficult to evaluate using conventional statistical methods. Another difficulty is identifying the cause of the failure to be able to intervene rapidly and in a focused manner in the event of irregularities. For this reason, this research study presented a comprehensive analysis of anomaly detection in binary time series data using various machine learning models. The study included preprocessing of the dataset, normalizing the data, and evaluation of the anomaly detection performance of the different models. The accuracy, detection rate, and F1-score are used as evaluation measures. The execution time of each model is also analyzed. In addition, the identification of sensors that cause anomalies is investigated and the impact of false detections is discussed. Experimental results show the strengths and weaknesses of each model and provide valuable insights for selecting the appropriate anomaly detection approach. The Isolation Forest, Local Outlier Factor, DBSCAN, and kMeans models show high precision and detection, while the Autoencoder and Variational Autoencoder models show high precision but lower detection. The one-class Support Vector Machine model achieves balanced performance. AutoML shows excellent results in recognition rate but is not real-time capable. The results highlight the trade-offs between performance and computational efficiency and point to further research potential in real-time implementation.

Learning Latent Representation with Limited Labels for IoT Anomaly Detection

Abstract— Malware detection is a critical challenge in the current era, especially for IoT devices. Previous studies have applied analytic techniques to reduce data size and extract valuable information. However, most of these studies count on a considerable quantity of outliers to perform anomaly detection. In this paper, we propose an enhanced method (named FeaWAD*) that improves the data encoding strategy based on the FeaWAD network [1]. These models require only a small fraction of anomalies for training. We evaluate the FeaWAD* method on the N-BaIoT dataset with various test scenarios for detecting known attacks as well as unknown future attacks. The experimental results demonstrate that the FeaWAD* method outperforms the original model FeaWAD and other popular anomaly detection methods such as Isolation Forest, Local Outlier Factor, and One-class Support Vector Machine. We also measure the time-based effectiveness of the model to assess its practical applicability.

Privileged multi-view one-class support vector machine

One-class support vector machine (OCSVM) is a typical one-class classification approach, which learns the classifier by using only the target samples. At present, most OCSVM works hypothesize that the samples have only one view, while multi-view OCSVM has not been taken into account. In this paper, a novel multi-view one-class support vector machine method with privileged information learning (MOCPIL) is put forward. MOCPIL embodies both the consensus principle and complementarity principle in multi-view learning. Privileged information is additional data that is available only in the training process, but not in the testing process. By introducing the idea of privileged information learning, MOCPIL implements the complementarity principle by treating one view as the training data and the other view as the privileged data. Moreover, MOCPIL implements the consensus principle by requiring that different views of the same object should give similar predicting outputs. The learning problem of MOCPIL is a quadratic programming (QP) problem, which is able to be solved by off-the-shelf QP solvers. To the best of our knowledge, this is the first study to tackle the multi-view learning problem based on OCSVM. The performance of MOCPIL is evaluated through extensive experiments. The experimental results have shown that MOCPIL explicitly outperforms the existing multi-view one-class classification methods.

The feasibility of using ATR-FTIR spectroscopy combined with one-class support vector machine in screening turmeric powders

Once turmeric has been ground into powder, it is difficult to tell visually if it has been tampered with. In this study, ATR-FTIR spectroscopy was used in tandem with one-class support vector machine (OCSVM) to detect adulteration in turmeric powder. The OCSVM models were trained using 42 pure turmeric powder samples, optimized using 30 pure turmeric powder samples, and subsequently evaluated by classifying 30 pure and 120 adulterated (cornstarch, Metanil Yellow, Orange II, and Sudan I) samples. Preprocessing methods, such as Savitzky-Golay (SG)-derivatives, standard normal variate (SNV), and multiplicative scatter correction (MSC), were used individually and in combination to obtain the best-performing model. Models were assessed by comparing the sensitivity, specificity, and efficiency values and compared with one-class soft independent modeling of class analogy (OCSIMCA). The best performing OCSVM model (sensitivity = 1.00, specificity = 0.89) was obtained by first conducting an MSC on the raw data followed by SG-2nd derivative transformation. It also has an efficiency value of 0.94, which was 0.14 higher than when data preprocessing was not done. Compared to the results of OCSIMCA, the OCSVM model gave a higher efficiency value and can detect lower levels of cornstarch adulteration. Also, the results showed that inclusion of data preprocessing can lead to a better classification model. With the obtained evaluation parameter values, ATR-spectroscopy coupled with OCSVM demonstrated its potential for screening turmeric powder products.

Open set recognition fault diagnosis framework based on convolutional prototype learning network for nuclear power plants

As an Open Set Recognition (OSR) problem, nuclear power plant fault diagnosis requires not only the correct classification of known faults, but also the effective identification of unknown faults, which cannot be satisfied by most previously developed models. In this study, a novel nuclear power plant OSR fault diagnosis framework based on Convolutional Prototype Learning (CPL) is proposed, in which CPL is used to extract discriminative fault features from raw nuclear power plant data. Besides, two classification methods, One-Class Support Vector Machine (OCSVM) method and Prototype Matching by Distance (PMD) method, are developed to complete the fault diagnosis of open space for the framework. To verify the feasibility and effectiveness of the proposed OSR framework, numerical experiments on 24 OSR tasks with high-dimensional and strong-nonlinear complex nuclear power plant simulation data are conducted, and the OSR performance evaluate by normalized accuracy and Youden's index, feature visualization and convergence rates are analyzed. Results show that compared with adopting traditional Convolutional Neural Network (CNN), the proposed CPL-based framework can significantly boost the OSR diagnostic performance on nuclear power plant fault diagnosis tasks, which benefits from the excellent ability of CPL to extract intra-class compact and inter-class separable feature representation.

A COMPARATIVE ANALYSIS OF ANOMALY DETECTION FROM MICROSERVICE GENERATED UNSTRUCTURED LOGS

In a process, to ensure increased reliability and better availability, it is very important to detect any anomalies that refer to any abnormality observed in the behaviour of a standard process. The breakdown of service(s) eventually leads to production loss, and at the same time, a system that is unreliable brings lots of challenges to the operations team. Anomaly detection plays a significant role to ensure that an application is reliable, secured and available for user requests. For the overall performance optimization of a cloud microservice based application without any disruption in service, and identification of possible security threat, it is much essential that the anomalies must be detected and responded to in time. In real life large microservice based production infrastructures environments, even though ample instance of normal activities is available, it is not possible to predict and create a dataset of anomalies. So these kind of data are not suitable for a supervised two-class classification. In this work, unsupervised one-class approaches such as Local Outlier Factor, Isolation Forest, and One Class SVM are used to find anomalies. On experimentation these models have obtained a high accuracy of 98% to 99%. On comparing the performance of the models, One-Class SVM is found to produce significantly higher number of False Positives in comparison to other two considered model

An Online Support Vector Machine Algorithm for Dynamic Social Network Monitoring

Online monitoring of social networks offers exciting features for platforms, enabling both technical and behavioral analysis. Numerous studies have explored the adaptation of traditional quality control methods for detecting change points within social networks. However, the current research studies face limitations such as an overreliance on case-based attributes, high computational costs, poor scalability with large networks, and low sensitivity in fast change point detection. This paper proposes a novel algorithm for social network monitoring using One-Class Support Vector Machines (OC-SVMs) to address these limitations. Additionally, using both nodal and network-level attributes makes it versatile for diverse social network applications and effectively detecting network disturbances. The algorithm utilizes a well-defined training data dictionary with an updating procedure for evolutionary networks, enhancing memory and time efficiency by reducing the processing of input data. Extensive numerical experiments are conducted using an EpiCNet model to simulate interactions in an online social network, covering six change scenarios to evaluate the proposed methodology. The results show lower Average Run Length (ARL) and Expected Delay Detection (EDD), demonstrating the superior accuracy and effectiveness of the OC-SVM algorithm compared to alternative methods. Applying OC-SVM to the Enron Email network indicates its capability to identify change points, reflecting the tumultuous timeline that led to Enron's downfall. This further validates the substantial advancement of OC-SVM in social network monitoring and opens doors to broader real-world applications.

Voltage-Induced Heating Defect Detection for Electrical Equipment in Thermal Images

Voltage-induced heating defect is a type of defect that may occur in transformation substation equipment. Although this type of defect is less common compared to current-induced heating defects, it is crucial to identify it due to its association with severe insulation degradation problems that require prompt intervention. However, the temperature variations caused by these defects may be relatively subtle, making it challenging to distinguish them in thermal images. In this work, considering the characteristics of voltage-induced heating defects and the scarcity of defect data, we propose a two-stage method for defect detection. In the first stage, we employ oriented R-CNN to detect oriented parts of the equipment, accurately localizing the centerline of each part. In the second stage, we extract the temperature distribution along the centerline of specific parts and discretize them as features. Subsequently, we train one-class support vector machines based on the features extracted from normal images for defect diagnosis. Experimental results demonstrate that the proposed method is capable of accurately detecting defects while maintaining a low false positive rate.

Conception and evaluation of anomaly detection models for monitoring analytical parameters in wastewater treatment plants

The exponential growth of technology in recent decades has led to the emergence of some challenges inherent to this growth. One of these challenges is the enormous amount of data collected by the different sensors in our society, namely in management processes such as Wastewater Treatment Plants (WWTPs). These infrastructures comprise several processes to treat wastewater and discharge clean water in water courses. Therefore, the concentration of pollutants must be below the allowable emissions limits. In this work, anomaly detection models were conceived, tuned and evaluated to monitor essential parameters such as nitrate and ammonia concentrations and pH to improve WWTP management. Four Machine Learning models were considered, particularly Local Outlier Fraction, Isolation Forest, One-Class Support Vector Machines and Long Short-Term Memory-Autoencoders (LSTM-AE), to detect anomalies in the three parameters mentioned. Through the different experiments, it was possible to verify that, in terms of F1-Score, the best candidate model for the three analyzed parameters was LSTM-AE-based, with a value consistently higher than 97%.

Prediction of Potato (Solanum tuberosum L.) Yield Based on Machine Learning Methods

This research delves into the application of machine learning methods for predicting the yield of potato varieties used for French fries in Poland. By integrating a comprehensive dataset comprising agronomical, climatic, soil, and satellite-based vegetation data from 36 commercial potato fields over five growing seasons (2018–2022), we developed three distinct models: non-satellite, satellite, and hybrid. The non-satellite model, relying on 85 features, excludes vegetation indices, whereas the satellite model includes these indices within its 128 features. The hybrid model, combining all available features, encompasses a total of 165 features, presenting the most-comprehensive approach. Our findings revealed that the hybrid model, particularly when enhanced with SVM outlier detection, exhibited superior performance with the lowest Mean Absolute Percentage Error (MAPE) of 5.85%, underscoring the effectiveness of integrating diverse data sources into agricultural yield prediction. In contrast, the non-satellite and satellite models displayed higher MAPE values, indicating less accuracy compared to the hybrid model. Advanced data-processing techniques such as PCA and outlier detection methods (LOF and One-Class SVM) played a pivotal role in model performance, optimising feature selection and dataset refinement. The study concluded that machine learning methods, particularly when leveraging a multifaceted approach involving a wide array of data sources and advanced processing techniques, can significantly enhance the accuracy of agricultural yield predictions. These insights pave the way for more-efficient and -informed agricultural practices, emphasising the potential of machine learning in revolutionising yield prediction and crop management.

Bounded exponential loss function based AdaBoost ensemble of OCSVMs

As a commonly used ensemble method, AdaBoost has drawn much consideration in the field of machine learning. However, AdaBoost is highly sensitive to outliers. The performance of AdaBoost may be greatly deteriorated when the training samples are polluted by outliers. For binary and multi-class classifications, there have emerged many approaches to improving the robustness of AdaBoost against outliers. Unfortunately, there are too few researches on enhancing the robustness of AdaBoost against outliers in the case of one-class classification. In this study, the exponential loss function of AdaBoost is replaced by a more robust one to improve the anti-outlier ability of the conventional AdaBoost based ensemble of one-class support vector machines (OCSVMs). Furthermore, based on the redesigned loss function, the update formulae for the weights of base classifiers and the probability distribution of training samples are reformulated towards the AdaBoost ensemble of OCSVMs. The empirical error upper bound is derived from the theoretical viewpoint. Experimental outcomes upon the artificial and benchmark data sets show that the presented ensemble approach is more robust against outliers than its related methods.