Underwater UXO classification using Matched Subspace Classifier with synthetic sparse dictionaries

This report describes the hardware system and the set of algorithms we have developed for detecting damage in cables for the Advanced Development and Process Technologies (ADAPT) Program. This program is part of the W80 Life Extension Program (LEP). The system could be generalized for application to other systems in the future. Critical cables can undergo various types of damage (e.g. short circuits, open circuits, punctures, compression) that manifest as changes in the dielectric/impedance properties of the cables. For our specific problem, only one end of the cable is accessible, and no exemplars of actual damage are available. This work addresses the detection of dielectric/impedance anomalies in transient time domain reflectometry (TDR) measurements on the cables. The approach is to interrogate the cable using time domain reflectometry (TDR) techniques, in which a known pulse is inserted into the cable, and reflections from the cable are measured. The key operating principle is that any important cable damage will manifest itself as an electrical impedance discontinuity that can be measured in the TDR response signal. Machine learning classification algorithms are effectively eliminated from consideration, because only a small number of cables is available for testing; so a sufficient sample size is not attainable. Nonetheless, a key requirement is to achieve very high probability of detection and very low probability of false alarm. The approach is to compare TDR signals from possibly damaged cables to signals or an empirical model derived from reference cables that are known to be undamaged. This requires that the TDR signals are reasonably repeatable from test to test on the same cable, and from cable to cable. Empirical studies show that the repeatability issue is the 'long pole in the tent' for damage detection, because it is has been difficult to achieve reasonable repeatability. This one factor dominated the project. The two-step model-based approach is summarized as follows: Step 1, Cable Modeling: Given input-output TDR signals s(n) and x{sub U}(n) for a cable known to be free of damage, system identification algorithms are used to compute a dynamic prediction-error cable model that has output {cflx x}{sub U}(n). The model is declared valid when the innovations e{sub U}(n) = x{sub U}(n) {cflx x}{sub U}(n) satisfy a statistical zero-mean whiteness test. This validated model output is then used as a known reference to which other cables can be compared. Step 2, Cable Testing: The TDR output signal x{sub D}(n) from a cable under test is compared with the model output {cflx x}{sub U}(n) by computing the innovations e{sub D}(n) = x{sub D}(n) {cflx x}{sub U}(n). The innovations are tested using a short-term whiteness test statistic, which employs a statistical confidence interval. If the cable passes the test, this implies that the model is valid and the cable is declared undamaged. If the cable fails the test, this indicates a model mismatch, which means that the cable's dielectric properties have changed; and this implies that the cable is damaged. The test threshold is adjusted to maximize probability of detection and minimize probability of false alarm according to an empirically determined receiver operating characteristic (ROC) curve. An associated confidence interval on the probability of correct classification is also provided. The effectiveness of the algorithm is demonstrated using measured TDR signals for undamaged and damaged cables. Experimental and algorithmic methods for coping with repeatability issues are presented. The model-based damage detection algorithms have been shown to perform well for some representative examples of real TDR signals acquired using the two-dimensional (2D) mockup fixture. If the damage causes a short circuit, then damage detection performance is generally good to excellent. Examples include the cases demonstrated in this report for cuts and pinholes. If the damage does not cause a short circuit, then damage detection performance is generally poor to fair. Examples include the cases demonstrated in this report for crushes and partial cuts.

Model-based analysis of production systems is one of the main areas in manufacturing research. The foundation of the successful application of these theoretical studies is the availability of valid and high-fidelity mathematical models that are capable of capturing the behavior of job flow in production systems. The modeling process of a production system, however, may require a significant amount of nonstandardized work that can only be done properly by someone with solid training in the area and extensive experience through real case studies. This poses a critical challenge in the effective implementation of these valuable theoretical results in the Industry 4.0 era. To overcome this, we propose a new production systems modeling paradigm inspired by system identification: calculate production system model parameters that best match the standard system performance metrics measured on the factory floor. Specifically, in this article, we consider production lines characterized by the Bernoulli serial line model and develop algorithms that identify model parameters to fit the system throughput and work-in-process. Analytical algorithms are derived to solve this problem in a two-machine line case and then extended to multi-machine lines. The accuracy and computational efficiency of the algorithms are demonstrated through extensive numerical experiments. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —A high-fidelity mathematical model is of critical importance to the implementation of any model-based production system analysis method. Currently, the construction of such models is carried out in an ad hoc manner. The quality of the resulting models may heavily depend on the training, experience, intuition, and personal preference of the modeler. The proposed model parameter identification method focuses on standard key performance indices commonly measured on the factory floor. The advantage is twofold. First, these standard performance metrics are consistently defined regardless of industry, thus avoiding any data-ambiguity issue that may occur when using complex machine/equipment status data. Second, measuring these performance metrics in real time is typically convenient and cost effective, even for manufacturing plants without high-end IT infrastructure, thus making the technology accessible to not only large but also small- and mid-sized manufacturers. Using the algorithms developed in this article, a practitioner can quickly construct a serial production line model and then utilize it to access the rich library of production analysis, design, and control methods available in the literature.

Much of the research into multipath detectionand mitigation has not considered the carrier phase delaybetween the line of sight (LOS) and reflected signals. Anew variable referred to as early late phase (ELP) hasrecently been proposed to exploit this phase difference. Ithas been found that in a receiver tracking the L1 GPSsignal, the probability of detecting multipath is lower whenthe carrier phase difference between the LOS and areflected signal is an integer multiple of p. Since thepseudorange error caused by the multipath’s presence is thehighest in this case, we propose to exploit the coexistenceof another GPS civilian signal, the L2C. We present ananalysis of ELP for the L1 and L2C signals, and a com-bination of both, for detecting multipath. The multipathdetection performance has been compared using probabil-ities of false alarm and detection. An ideal algorithmshould have lower probability of false alarm and higherprobability of detection. However, it has been found thatusing dual-frequency ELP increases both probabilities.Thus, receiver operator characteristics (ROC) curves, andthe area under the ROC curves, have been used for effec-tive comparison. It has been found that the L2C signalindividually gives worse performance than L1 because ofits weaker signal strength. However, the combination of L1and L2C gives the best overall performance, and thus it canbe claimed that ELP using dual-frequency receivers is amore effective approach for detecting multipath.Keywords Multipath GPS L2C ELP Dual-frequency receiverIntroductionThe last decade or so has seen an increase in applicationsof GPS in urban and indoor applications, which hasencouraged extensive research activity in the field ofmultipath detection and mitigation. Algorithms widelycited are narrow correlators (Dierendonck et al. 1992),double-difference residuals (DDR) (Townsend and Fenton1994), the vision correlator (Fenton and Jones 2005) andthe multipath estimating delay locked loop (MEDLL)(Townsend et al. 2000). Since a reflected signal reachesthe receiver by a relatively longer path, it is delayed intime and also has a carrier phase difference with respectto the line of sight (LOS) signal (Hatch et al. 2007;Mubarak and Dempster 2008). The above algorithms (andmany others) only use the magnitude of the correlatoroutputs and hence do not exploit this phase difference. Anew parameter, called the ‘‘early late phase (ELP)’’, hasrecently been proposed for multipath detection and esti-mation (Mubarak and Dempster 2007). ELP is the phasedifference between the early and late correlator outputs,where the phase of a correlator output is equal to theinverse tangent of the Q-channel output divided by theI-channel output. It is higher in the presence of multipathand has been shown to be a useful discriminator to detectthe presence of multipath (Mubarak 2008).This paper first provides a brief description of ELP,followed by its analysis in the presence of multipath for theL1 and L2C signals, and the combination of both. Multi-path detection performance is then compared using prob-abilities of false alarm and detection. ROC curves and the

Heart rate variability (HRV) analysis is considered as a preliminary diagnosis method to check the cardiac health of the human heart. The reliability of the HRV analysis system solely depends on the accuracy of the QRS complex detector. Hence, in this paper, an optimally designed digital differentiator (DD) for precise detection of QRS complex is proposed. The proposed DD is designed by using an efficient evolutionary optimization technique called gases Brownian motion optimization (GBMO) algorithm and is used in the preprocessing stage of the QRS detector. In GBMO algorithm, a balanced trade-off is maintained between both the exploration and the exploitation phases to find the global optimum solution. The electrocardiogram signal is preprocessed by using the proposed DD to generate the feature signals corresponding to the R-peaks only. The detection technique utilizes the principle of Hilbert transform and zeroes crossing detection. The proposed approach is verified against all the first channel records of MIT/BIH arrhythmia database by considering the standard QRS detection performance metrics and produces a sensitivity (Se) of 99.92%, positive predictivity (+P) of 99.92%, detection error rate (DER) of 0.1562%, QRS detection rate of 99.92%, accuracy (Acc) of 99.84%, and F score of 0.9992%. With respect to the standard performance metrics, the proposed QRS detector outperforms all the recently reported QRS detection techniques.

Receiver operating characteristic curve analysis of clinical risk models

We sought to identify factors that are associated with hypotension after the induction of spinal anesthesia (SpA) by using an anesthesia information management system. Hypotension was defined as a decrease of mean arterial blood pressure of more than 30% within a 10-min interval, and relevance was defined as a therapeutic intervention with fluids or pressors within 20 min. From January 1, 1997, to August 5, 2000, data sets from 3315 patients receiving SpA were recorded on-line by using the automatic anesthesia record keeping system NarkoData. Hypotension meeting the predefined criteria occurred in 166 (5.4%) patients. Twenty-nine patient-, surgery-, and anesthesia-related variables were studied by using univariate analysis for a possible association with the occurrence of hypotension after SpA. Logistic regression with a forward stepwise algorithm was performed to identify independent variables (P < 0.05). The discriminative power of the logistic regression model was checked with a receiver operating characteristic curve. Calibration was tested with the Hosmer-Lemeshow goodness-of-fit test. The univariate analysis identified the following variables to be associated with hypotension after SpA: age, weight, height, body mass index, amount of plain bupivacaine 0.5% used for SpA, amount of colloid infusion before puncture, chronic alcohol consumption, ASA physical status, history of hypertension, urgency of surgery, surgical department, sensory block height of anesthesia, and frequency of puncture. In the multivariate analysis, independent factors for relevant hypotension after SpA consisted of three patient-related variables ("chronic alcohol consumption," odds ratio [OR] = 3.05; "history of hypertension," OR = 2.21; and the metric variable "body mass index," OR = 1.08) and two anesthesia-related variables ("sensory block height," OR = 2.32; and "urgency of surgery," OR = 2.84). The area of 0.68 (95% confidence interval, 0.63-0.72) below the receiver operating characteristic curve was significantly greater than 0.5 (P < 0.01). The goodness-of-fit test showed a good calibration of the model (H = 4.3, df = 7, P = 0.7; C = 7.3, df = 8, P = 0.51). This study contributes to the identification of patients with a high risk for hypotension after SpA induction, with the risk increasing two- or threefold with each additional risk factor. By using automated data collection, 5 (chronic alcohol consumption, history of hypertension, body mass index, sensory block height, and urgency of surgery) of 29 variables could be detected as having an association with hypotension after spinal anesthesia induction. The knowledge of these risk factors should be useful in increasing vigilance in those patients most at risk for hypotension, in allowing a more timely therapeutic intervention, or even in suggesting the use of alternative methods of spinal anesthesia, such as titrated continuous or small-dose spinal anesthesia.

This research paper explores the use of machine learning techniques to classify underwater objects as rocks or mines using SONAR echo data.Accurate detection of underwater mines is essential for maritime safety and ocean engineering, as traditional manual methods often lack efficiency and reliability.In this study, supervised machine learning models are applied to identify distinguishing patterns in SONAR signal features that separate natural rock formations from man-made mines.The dataset is pre-processed using normalization and feature standardization to improve learning performance and prediction accuracy.Classification algorithms such as Logistic Regression and Random Forest are trained and evaluated using standard performance metrics including accuracy, precision, recall, F1-score, and ROC-AUC.The experimental results show that the implemented models achieve reliable and consistent classification outcomes.This study highlights the effectiveness of machine learning-based analysis for automated underwater object detection and demonstrates its potential in reducing human intervention.The proposed approach provides a basis for developing intelligent detection systems and can be further enhanced by incorporating advanced learning methods and real-time SONAR data.

Performance metrics are measures of success or performance that can be used to evaluate how well a model makes accurate predictions or classifications. However, there is no single measure since each performance metric emphasizes a different classification aspect. Model selection procedures based on information criteria offer a quantitative measure that balances model complexity with goodness of fit, providing a better alternative to classical approaches. In this paper, we introduce and develop a novel Information Complexity-Receiver Operating Characteristic, abbreviated as ICOMP-ROC, criterion approach to fit and study the performance of ROC curve models. We construct and derive the Universal ROC (UROC) for a combination of sixteen Bi-distributional ROC models to choose the best Bi-distributional ROC by minimizing the ICOMP-ROC criterion. We conduct large-scale Monte Carlo simulations using the sixteen Bi-distributional ROC models with the Normal-Normal and Weibull-Gamma pairs as the pseudo-true ROC models. We report the frequency of hits of the ICOMP-ROC criterion, showing its remarkable recovery rate. In addition to Bi-distributional fitting, we consider a high-dimensional real Magnetic Resonance Imaging (MRI) of the Brain dataset and Wisconsin Breast Cancer (WBC) dataset to study the performance of the common performance metrics and the ICOMP-ROC criterion using several machine learning (ML) classification algorithms. We use the genetic algorithm (GA) to reduce the dimensions of these two datasets to choose the best subset of the features to study and compare the performance of the newly proposed ICOMP-ROC criterion along with the traditional performance metrics. The choice of a suitable metric is not just contingent upon the ML model used, but it also depends upon the complexity and high dimensionality of the input datasets, since the traditional performance metrics give different results and have inherent limitations. Our numerical results show the consistency and reliability of the ICOMP-ROC criterion over the traditional performance metrics as a clever model selection criterion to choose the best fitting Bi-distributional ROC model and the best classification algorithm among the ones considered. This shows the utility and the versatility of our newly proposed approach in ROC curve modeling that integrates and robustifies currently used procedures.

Ensuring high-quality production of glass products is essential in modern manufacturing, where even minor defects such as scratches, cracks, or surface irregularities can significantly affect safety, performance, and usability. Traditional visual inspection methods are labour-intensive, time-consuming, and prone to human error, creating a strong need for automated and reliable defect detection systems. The main objective of this research is to develop an efficient Deep Learning (DL) based framework for accurate binary classification of defective and non-defective glass products while addressing limitations such as subtle defect visibility, variation in lighting conditions, and dataset imbalance. Furthermore, the study aims to evaluate model performance under different hyperparameter settings using an intelligent optimisation technique. The proposed work employs the Residual Network 50 Version 2 (ResNet50V2) Convolutional Neural Network (CNN) for defect classification, supported by a comprehensive preprocessing pipeline consisting of resizing, normalisation, greyscale conversion, and data augmentation. Bayesian Optimisation (BO) is integrated to fine-tune key hyperparameters, ensuring optimal learning efficiency and improved generalisation. The framework is trained using a curated glass defect dataset and evaluated using standard performance metrics to demonstrate its robustness and suitability for real-world industrial inspection scenarios. Experimental results show that the optimised ResNet50V2-BO model achieves 98.36% accuracy, along with a precision of 98%, recall of 93%, and F1-score of 95% for the defective class. The confusion matrix and PR curves further confirm the model’s ability to reliably distinguish subtle surface defects. These findings highlight the potential of the proposed approach as a fast, reliable, and scalable solution for automated quality control in glass manufacturing industries.

Abstract. This paper presents a data-driven framework to assess the impact of green hydrogen integration on the voltage stability of a part of the Moroccan electrical transmission network. The main contribution lies in modeling green hydrogen as a long-duration energy storage system capable of absorbing surplus solar and wind generation and reinjecting it into the grid to enhance voltage stability under high renewable penetration. A time-series power flow methodology, implemented in Python using the pandapower library, is developed to generate a large dataset based on voltage criteria. This dataset is then used to train and evaluate four machine learning models (RBF-SVM, Random Forest, QDA and Naive Bayes) through confusion matrices and standard performance metrics. The comparative results indicate that the RBF-SVM model clearly outperforms the other classifiers across all evaluation metrics, including accuracy, precision, recall and F1-score. Beyond model performance, the proposed approach introduces an integrated framework that combines hydrogen-based long-term energy storage modeling with machine learning–based stability prediction.

Introduction: We propose a novel approach for the assessment of the similarity of retinal vessel segmentation images that is based on linking the standard performance metrics of a segmentation algorithm, with the actual structural properties of the images through the fractal dimension.Method: We apply our methodology to compare the vascularity extracted by automatic segmentation against manually segmented images.Results: We demonstrate that the strong correlation between the standard metrics and fractal dimension is preserved regardless of the size of the subimages analyzed.Discussion or Conclusion: We show that the fractal dimension is correlated to the segmentation algorithm’s performance and therefore it can be used as a comparison metric.

Receiver Operating Characteristic (ROC) curves can be used to describe detectability of noise-limited signals. The basic concepts underlying the theory of ROC curves and the experimental determination of ROC curves are discussed. ROC curves are used to describe the detectability of the image of 2 mm Lucite beads (similar to certain kinds of small gall-stones) in a noisy background of radiographic mottle. Results are shown for four typical radiographic screen-film combinations of differing optical and noise properties. The qualitative relationship between the ROC curves and the mathematical model describing radiographic mottle is examined. The ROC curves describing the detectability of the bad images when radiographed with two screen-film combinations of equal system speed but different optical characteristics are compared to the quality of blood vessel images obtained with the same screen-film combinations. It is concluded that it is important to distinguish between the assessment of image quality based on the relative sharpness of high contrast images, versus the assessment of image quality based on the relative detectability of low contrast noise-limited images.

295: Accuracy of Various Methods of B-Line Assessment In the Diagnosis of Congestive Heart Failure Compared by Receiver Operating Characteristic Curves

Buried unexploded ordnance (UXO) and land mines are grave threats to civilians who go back to contaminated areas in the postwar period. In this research, we propose a new technique to detect UXOs from ground penetrating radar (GPR) signals. The technique is based on an energy-based feature of the region of interest (ROI). A ROI selection technique in range domain is also proposed. The proposed method was tested with the UXO data collected by Battelle company and the Ohio State University. This data set was made available to public through the Unexploded Ordnance Center of Excellence (UXOCOE), Department of Defense, U.S.A. The results evaluated in terms of Receiver Operating Characteristic (ROC) curves suggest that our proposed technique performs very well. Comparisons to a traditional detection technique and our previously proposed technique are conducted. The improvement by the proposed technique does not only speed up the UXO/land mine clearance operation, but also save operating cost.

This research compares alternative performance metrics to those more commonly used in target detection system performance evaluation. The alternative performance metrics examined here include the Fisher ratio, a modification of the Dice similarity coefficient, and the Youden index. These metrics are compared to metrics that have been previously introduced for such target detection system performance evaluation: the receiver operating characteristic (ROC) curve (and the related summary area under the ROC curve (AUC) value), and the confidence error generation (CEG) curve (and the related summary root square deviation (RSD) value). The ROC curve is a discrimination metric that measures the ability of a target detection system to distinguish between target and non-target. The CEG curve quantifies detection system knowledge of its own performance. An approach is presented that combines such metrics; this combination may be dynamically adjusted and updated based on current and future evaluation requirements for particular target detection systems.