Performance Analysis Of Detection Research Articles

Classification of HHO-based Machine Learning Techniques for Clone Attack Detection in WSN

Thanks to recent technological advancements, low-cost sensors with dispensation and communication capabilities are now feasible. As an example, a Wireless Sensor Network (WSN) is a network in which the nodes are mobile computers that exchange data with one another over wireless connections rather than relying on a central server. These inexpensive sensor nodes are particularly vulnerable to a clone node or replication assault because of their limited processing power, memory, battery life, and absence of tamper-resistant hardware. Once an attacker compromises a sensor node, they can create many copies of it elsewhere in the network that share the same ID. This would give the attacker complete internal control of the network, allowing them to mimic the genuine nodes' behavior. This is why scientists are so intent on developing better clone assault detection procedures. This research proposes a machine learning based clone node detection (ML-CND) technique to identify clone nodes in wireless networks. The goal is to identify clones effectively enough to prevent cloning attacks from happening in the first place. Use a low-cost identity verification process to identify clones in specific locations as well as around the globe. Using the Optimized Extreme Learning Machine (OELM), with kernels of ELM ideally determined through the Horse Herd Metaheuristic Optimization Algorithm (HHO), this technique safeguards the network from node identity replicas. Using the node identity replicas, the most reliable transmission path may be selected. The procedure is meant to be used to retrieve data from a network node. The simulation result demonstrates the performance analysis of several factors, including sensitivity, specificity, recall, and detection.

Acousto-Ultrasonic Inspection of Pre-baked Carbon Anodes: Repeatability Analysis and Assessment of Damage Detection Performance

Acousto-Ultrasonic Inspection of Pre-baked Carbon Anodes: Repeatability Analysis and Assessment of Damage Detection Performance

Swish and gargle saliva sampling is a patient-friendly and comparable alternative to nasopharyngeal swabs to detect SARS-CoV-2 in outpatient settings for adults and children.

Nasopharyngeal swabs (NPS) are considered the standard specimen for SARS-CoV-2 detection by PCR. We aimed to compare detection performance and patient experience of NPS with swish and gargle saliva (SGS) to determine its potential as an accurate, easy-to-collect, and comfortable alternative. We conducted a prospective study from March to May 2021 and recruited pediatric and adult outpatients seeking COVID-19 testing. Paired NPS and 5-mL SGS were collected from each participant during the same visit and tested in parallel using validated rRT-PCR assays. Variant of concern (VOC) analysis was performed for positive NPS specimens. The participant completed the surveys regarding their experience with the two collection methods. We included 238 participants in the SARS-CoV-2 detection performance analysis. Thirty-two participants tested positive for both NPS and SGS. In comparison to NPS, the sensitivity and specificity of SGS to detect SARS-CoV-2 were 94.1% (95%CI; 80.3%-99.3%) and 97.1% (95%CI; 93.7%-98.9%), respectively. Six participants tested positive only with SGS and only two with NPS. NPS is an imperfect gold standard. Thus, when compared to the COVID-19 case, that is, a patient with positive NPS and/or SGS, the sensitivity for NPS and SGS were 85.0% and 95.0%, respectively. VOC results (n = 26) revealed that 92% of the cases were alpha. From the 238 surveys, 89.9% of the participants described SGS collection as comfortable or very comfortable compared to 15.1% for NPS; 90.2% of the participants were likely or very likely to return for SGS collection compared to 59.3% for NPS. SGS is an alternative to NPS for SARS-CoV-2 detection in adult and pediatric outpatients as it is more patient-friendly while still maintaining comparable performance.IMPORTANCEWidespread and frequent testing for COVID-19 was an important strategy to identify infected patients to isolate and control the spread of the disease during the pandemic. The nasopharyngeal swab (NPS) global supply chain and access to trained healthcare professionals for standard NPS collection were often compromised. Patient discomfort and limited access challenged health systems to reach large numbers for testing in adult and pediatric populations. Our study revealed that swish and gargle saliva (SGS) was comparable to NPS in detecting SARS-CoV-2 and more patient-friendly than NPS. Patients were more likely to repeat the test with SGS. SGS was amenable to self-collection instead of relying on skilled professionals. This comprehensive evaluation highlights the challenges of comparing the accuracy of new methods to imperfect gold standards and identifies additional patient-centric factors that should be considered when defining such standards. Thus, SGS is an advantageous alternative specimen collection for outpatient en masse testing.

Performance Analysis of Deep Transfer Learning Models for the Automated Detection of Cotton Plant Diseases

Cotton is one of the most important agricultural products and is closely linked to the economic development of Pakistan. However, the cotton plant is susceptible to bacterial and viral diseases that can quickly spread and damage plants and ultimately affect the cotton yield. The automated and early detection of affected plants can significantly reduce the potential spread of the disease. This paper presents the implementation and performance analysis of bacterial blight and curl virus disease detection in cotton crops through deep learning techniques. The automated disease detection is performed through transfer learning of six pre-trained deep learning models, namely DenseNet121, DenseNet169, MobileNetV2, ResNet50V2, VGG16, and VGG19. A total of 1362 images of local agricultural fields and 1292 images from online resources were used to train and validate the models. Image augmentation techniques were performed to increase the dataset diversity and size. Transfer learning was implemented for different image resolutions ranging from 32×32 to 256×256 pixels. Performance metrics such as accuracy, precision, recall, F1 Score, and prediction time were evaluated for each implemented model. The results indicate higher accuracy, up to 96%, for DenseNet169 and ResNet50V2 models when trained on the 256×256 pixels image dataset. The lowest accuracy, 52%, was obtained by the MobileNetV2 model when trained on low-resolution, 32×32, images. The confusion matrix analysis indicates the true-positive prediction rates higher than 91% for fresh leaves, 87% for bacterial blight, and 76% for curl virus detection for all implemented models when trained and tested on an image dataset of 128×128 pixels or higher resolution.

The performance analysis of the laser heterodyne ultra-high frequency detection based on the optical phase-locked loop

The performance of the laser heterodyne ultra-high frequency detection based on the optical phase-locked loop is analyzed through theoretical calculations and numerical simulations, considering the residual phase error and carrier stability. Appling the Wiener–Khinchin theorem, a general expression of the residual phase variance with essential affecting factors is derived. Furthermore, we define the error gain factor based on the ultra-high frequency carrier stability, which consists of the amplitude deviation and frequency deviation. Then the model of the demodulation output with the amplitude deviation and frequency deviation is presented. According to numerical simulation results, we acquired the variation curve of the residual phase variance with the loop bandwidth and the performance of the carrier signal stability. For the residual phase variance expression, the minimum variance corresponding to the optimal parameters can be obtained through the numerical simulation. The simulation experiment proves that the numerical results are consistent with the theoretical analysis. Our model for evaluating the influence of the carrier stability on the heterodyne ultra-high frequency detection can provide powerful support for improving the demodulation accuracy. The quantitative analysis of the paper will be available for theoretical basis and guidance for laser heterodyne detection based on optical phase-locked loops.

A Covert Jamming Scheme Against an Intelligent Eavesdropper in Cooperative Cognitive Radio Networks

In this work, we design a covert jamming scheme against an intelligent Eve towards physical layer security in a cooperative cognitive radio networks. To protect the primary message from being decoded by an intelligent eavesdropper (Eve), a secondary user is picked as a friendly jammer that transmits artificial noise (AN) to the Eve. However, the intelligent Eve can detect the existence of AN, and thus the AN is inclined to be covertly transmitted at a lower detection probability, thereby mistaking the AN as background noise by the Eve. In this paper, firstly, by using a binary hypothesis testing, the detection error probability (DEP) is explicitly derived for detection performance analysis of the Eve in the perfect as well as statistical channel state information (CSI) scenario, separately. With the DEP, a covert jamming scheme is formulated to transmit the AN in both cases. Then, an optimization problem is formulated, and then recast into a two-part problem to maximize the secrecy performance with the covert jamming scheme in the perfect CSI case. Finally, the effective secrecy throughput is maximized under convert constraints in the statistical CSI case. Numerical results are provided to validate the DEP at the Eve and secrecy performance analysis.

Cyber security: performance analysis and challenges for cyber attacks detection

Nowadays, with the occurrence of new attacks and raised challenges have been facing the security of computer systems. Cyber security techniques have become essential for information technology services to detect and react against cyber-attacks. The strategy of cyber security enables visibility of various types of attacks and vulnerabilities throughout computer networks, whilst also provides detecting cyber-attacks and effective ways of identifying and preventing them. This study mainly focuses on the performance analysis and challenges faced by cyber security using the latest techniques. It also provides a review of the attack detection process including the robust effectiveness of intelligent techniques. Finally, summarize and discuss some methods to increase attack detection performance utilizing deep learning (DL) architectures.

Wavefront detection performance analysis of plenoptic sensor

Wavefront detection performance analysis of plenoptic sensor

협력 인지 무선 네트워크 양자화 비트 수 할당기법의 감지성능분석

협력 인지 무선 네트워크 양자화 비트 수 할당기법의 감지성능분석

PROMETHEE-Based Multi-AUV Threat Assessment Method Using Combinational Weights

The assessment of multiple incoming autonomous underwater vehicles (multi-AUVs) and threat prioritization are critical to underwater defense. To solve problems troubling multi-AUV threat assessment solutions, such as difficult data analysis, high subjectivity, and rigid prioritization logic, we propose the PROMETHEE algorithm based on fusion weights calculated twice by entropy and an analytic network process (ANP), respectively. First, according to AUV detection performance and underwater confrontation situation analysis, the main criteria and indicators of threat assessment are determined. The threat assessment system is provided by unified measurement of these indicators. Then, through analysis and assessment, the weighting algorithm is designed using entropy and ANP. The subjective weight calculated based on ANP and the objective weight obtained based on the entropy method are fused twice to obtain the combined weights, and the influence of subjective and objective factors on problem analysis is considered. Finally, by analyzing the simulation results of a multi-AUV, it is proven that the proposed algorithm is scientific and effective for AUV threat assessment. According to the experimental results, accurate evaluation of the target improved by at least 10%, enabling delivery of results close to the real confrontation situation with high reliability.

Analysis of Line and Tube Detection Performance of a Chest X-ray Deep Learning Model to Evaluate Hidden Stratification.

This retrospective case-control study evaluated the diagnostic performance of a commercially available chest radiography deep convolutional neural network (DCNN) in identifying the presence and position of central venous catheters, enteric tubes, and endotracheal tubes, in addition to a subgroup analysis of different types of lines/tubes. A held-out test dataset of 2568 studies was sourced from community radiology clinics and hospitals in Australia and the USA, and was then ground-truth labelled for the presence, position, and type of line or tube from the consensus of a thoracic specialist radiologist and an intensive care clinician. DCNN model performance for identifying and assessing the positioning of central venous catheters, enteric tubes, and endotracheal tubes over the entire dataset, as well as within each subgroup, was evaluated. The area under the receiver operating characteristic curve (AUC) was assessed. The DCNN algorithm displayed high performance in detecting the presence of lines and tubes in the test dataset with AUCs > 0.99, and good position classification performance over a subpopulation of ground truth positive cases with AUCs of 0.86-0.91. The subgroup analysis showed that model performance was robust across the various subtypes of lines or tubes, although position classification performance of peripherally inserted central catheters was relatively lower. Our findings indicated that the DCNN algorithm performed well in the detection and position classification of lines and tubes, supporting its use as an assistant for clinicians. Further work is required to evaluate performance in rarer scenarios, as well as in less common subgroups.

Longitudinal Analysis of Early Hearing Detection and Intervention Program Performance.

Early hearing detection and intervention (EHDI) is a newborn hearing screening system created to detect infants with hearing loss (HL) and intervene to reduce language and communication impairment. Early hearing detection (EHD) consists of three sequential stages: identification, screening, and diagnostic testing. This study longitudinally reviews each stage of EHD in each state and proposes a framework to improve utilization of EHD data. A retrospective public database review was conducted, accessing publicly available data from the Centers for Disease Control and Prevention. Summary descriptive statistics were utilized to generate a descriptive study of EHDI programs in each U.S. state from 2007 to 2016. Data over 10 years from 50 states as well as Washington, DC were included in this analysis, creating up to 510 data points per analysis. Hundred percent (85 to 105) (median [min to max]) of newborns were identified by and entered EHDI programs. Ninety-eight percent (51 to 100) of identified infants completed screening. Of the infants who screened positive for HL, the proportion that received diagnostic testing was 55% (1 to 100). The overall proportion of infants who failed to complete EHD was 3% (1 to 51). Of the infants who fail to complete EHD 70% (0 to 100) are from missed screenings, 24% (0 to 95) are from missed diagnostic testing, and 0% (0 to 93) are from missed identification. Although there are more infants missed at screening, it was estimated, with limitations, that there is an order of magnitude more infants with HL among those who did not complete diagnostic testing compared with those who did not complete screening. Analysis demonstrates high completion rates at both identification and screening stages, whereas the diagnostic testing stage demonstrates low and highly variable completion rates. The low completion rates at diagnostic testing create a bottleneck in the EHD process and the large variability impedes the comparison of HL outcomes across states. Analysis also demonstrates that among all stages of EHD, whereas the largest number of infants are missed at screening, the largest number of children with HL are likely missed at diagnostic testing. Therefore, a focus by individual EHDI programs on addressing causes of low diagnostic testing completion rates would yield the greatest increase in the identification of children with HL. Potential causes of low diagnostic testing completion rates are further discussed. Finally, a new vocabulary framework is proposed to facilitate further study of EHD outcomes.

Machine Learning Based Performance Analysis of Video Object Detection and Classification Using Modified Yolov3 and Mobilenet Algorithm

Detecting foreground objects in video is crucial in various machine vision applications and computerized video surveillance technologies. Object tracking and detection are essential in object identification, surveillance, and navigation approaches. Object detection is the technique of differentiating between background and foreground features in a photograph. Recent improvements in vision systems, including distributed smart cameras, have inspired researchers to develop enhanced machine vision applications for embedded systems. The efficiency of featured object detection algorithms declines as dynamic video data increases as contrasted to conventional object detection methods. Moving subjects that are blurred, fast-moving objects, backdrop occlusion, or dynamic background shifts within the foreground area of a video frame can all cause problems. These challenges result in insufficient prominence detection. This work develops a deep-learning model to overcome this issue. For object detection, a novel method utilizing YOLOv3 and MobileNet was built. First, rather than picking predefined feature maps in the conventional YOLOv3 architecture, the technique for determining feature maps in the MobileNet is optimized based on examining the receptive fields. This work focuses on three primary processes: object detection, recognition, and classification, to classify moving objects before shared features. Compared to existing algorithms, experimental findings on public datasets and our dataset reveal that the suggested approach achieves 99% correct classification accuracy for urban settings with moving objects. Experiments reveal that the suggested model beats existing cutting-edge models by speed and computation.

Performance analysis of entropy variation-based detection of DDoS attacks in IoT

Internet of Things (IoT) comprises of connected devices that collaborate to perform numerous tasks individually and collectively to facilitate services to the users. Distributed denial of service attacks aim to exhaust the resources of the target by sending enormous inessential traffic from numerous sources. IoT systems, generally being a resource-constrained network, are highly vulnerable to DDoS attacks. We aim to achieve the detection of DDoS attacks by rule-based detection method using information theory metrics. Variation in entropy of traffic features beyond a threshold is an indicator of change in density of traffic to a network. In this paper, we present a performance analysis of different parameters associated with entropy variation-based detection of DDoS attacks. Three feature sets were used for studying the impact of entropy variation in detection. Further, we have analysed the three kinds of entropies used for the detection namely, Shannon entropy, generalized entropy, and ϕ-entropy. Along with that, threshold determination for entropy values have been done in two ways, i.e. [average−standarddeviation,average+standarddeviation] and [lower quartile, upper quartile]. We found that quartile-based threshold has generally given better recall and accuracy and more consistent results with respect to the window sizes. It is also found that ϕ-entropy has given overall better results as compared to the Shannon entropy and Generalized entropy. Therefore, a combination of ϕ-entropy and quartile-based threshold is found to be the best parameters to identify the DDoS attacks.

HQDCNet: Hybrid Quantum Dilated Convolution Neural Network for detecting covid-19 in the context of Big Data Analytics.

Medical care services are changing to address problems with the development of big data frameworks as a result of the widespread use of big data analytics. Covid illness has recently been one of the leading causes of death in people. Since then, related input chest X-ray image for diagnosing COVID illness have been enhanced by diagnostic tools. Big data technological breakthroughs provide a fantastic option for reducing contagious Covid disease. To increase the model's confidence, it is necessary to integrate a large number of training sets, however handling the data may be difficult. With the development of big data technology, a unique method to identify and categorise covid illness is now found in this research. In order to manage incoming big data, a massive volume of chest x-ray images is gathered and analysed using a distributed computing server built on the Hadoop framework. In order to group identical groups in the input x-ray images, which in turn segments the dominating portions of an image, the fuzzy empowered weighted k-means algorithm is then employed. A hybrid quantum dilated convolution neural network is suggested to classify various kinds of covid instances, and a Black Widow-based Moth Flame is also shown to improve the performance of the classifier pattern. The performance analysis of COVID-19 detection makes use of the COVID-19 radiography dataset. The suggested HQDCNet approach has an accuracy of 99.01. The experimental results are evaluated in Python using performance metrics such as accuracy, precision, recall, f-measure, and loss function.

Performance analysis of diabetic retinopathy detection using fuzzy entropy multi-level thresholding

Diabetic Retinopathy (DR) is one of the major causes of blindness. Many DR detection systems were developed to segment and determine the type and number of lesions that appeared on retinal images and used to classify DR and its severity level. Even though several researchers have already proposed many automated diagnosis systems with different image segmentation algorithms, their accuracy and reliability are generally unexplored. The accuracy of an automated diagnosis system usually depends on the segmentation techniques. The accuracy of this system is heavily dependent upon the retinal and image parameters, which have intensity level difference between background (BG)-blood vessels (BV), BV-bright lesions, BV-dark lesions, and noise levels. In this work, the automated diagnosis system accuracy has been analysed to successfully detect DR and its severity levels. The focus is on fundus image modalities segmentation based on fuzzy entropy multi-level thresholding. The analysis aimed to develop conditions to guarantee accurate DR detection and its severity level. Firstly, a retinal image model was developed that represents the retina under the variation of all retinal and image parameters. Overall, 45,000 images were developed using the retinal model. Secondly, feasibility and consistency analysis were performed based on a specific design Monte Carlo statistical method to quantify the successful detection of DR and its severity levels. The conditions to guarantee accurate DR detections are: BG to BV > 30% and BV to the dark lesions (MAs) >15% for mild DR, BG to BV > 40% and BV to the dark lesions (MAs and HEM) > 20% for moderate DR, and BG to BV > 30% and BV to the dark lesions (MAs and HEM) > 15%, and BV to the bright lesions (EX) > 55% for severe DR. Finally, the validity of these conditions was verified by comparing their accuracy against real retinal images from publicly available datasets. The verification results demonstrated that the condition for the analysis could be used to predict the success of DR detection.

Performance analysis of linear detection for uplink massive MIMO system based on spectral and energy efficiency with Rayleigh fading channels in 3D plotting pattern

AbstractMassive multiple‐input multiple‐output (MIMO) is a critical component of 5G cellular networks, which utilizes large numbers of antennas at both the transmitter and receiver to enhance throughput and radiated energy efficiency. Various linear detection techniques are employed with massive MIMO to counteract path loss and interference, and maximize throughput. The first aim of this paper is to analyse the performance of uplink massive MIMO system for different linear detection techniques including: Maximum ratio combining (MRC), zero‐forcing (ZF), regularized ZF (RZF) and minimum mean squared error (MMSE) over Rayleigh channel model. The second aim is to jointly investigate the optimal values of signal‐to‐noise ratio (SNR), the number of antennas M and the number of users K for maximizing the spectral efficiency (SE) and energy efficiency (EE) through simulation using MATLAB and 3D plotting patterns. The obtained results show that the best SE and EE are achieved by uplink massive MIMO setup while using optimal values of SNR, M and K. It is observed that MMSE achieved the best performance. However, it requires estimation of average SNR at BS. Therefore, the best choice is ZF or RZF without any need for SNR estimation.

Performance Analysis of Classification and Detection of Brain Tumor MRI Images Using Resnet50 Deep Residual U-Net

Background: This research adds to the growing body of work demonstrating the vital role of image categorization in the medical sector. The efficiency of illness diagnosis is being greatly enhanced using image classification. A brain tumor is a collection of abnormal cells in the brain. Diagnostic precision is required when looking for a tumor in the brain because even the smallest error in human judgement can have disastrous results. To get around this problem, the medical community has implemented an automated brain tumor segmentation system. A variety of methods are employed to segment a brain tumor, but the results are inconsistent. To improve the quality of MRI images, we present our findings in this paper. Deep learning methods for image segmentation and classification are discussed in this paper.
 Methods: In this paper we a very robust deep learning method for image segmentation and classification. For image classification, we are employing the enhanced faster R-CNN method. The ResNet50 model is used to differentiate between tumor and non-tumor images. The next step in this framework is to use Deep Residual UNET for segmentation. RESUNET is an FCNN that maximizes efficiency. The proposed method works well in terms of its ability to find and classify things accurately.
 Results: The accuracy rate for identifying tumours in MRI scans using the proposed technique is 94.23%. Using transfer learning with Resnet50 as the base model and the use of discriminative learning rates, our model achieved superior results than other recent models.
 Conclusion: Within the scope of this study, we have integrated the residual networks with the U-Net to make the network stronger. This strategy improves the efficiency of classification and segmentation tools. To achieve a higher level of accuracy, the model may be trained further, or additional data may be applied in the training process.

Performance Analysis of Classification and Detection for PV Panel Motion Blur Images Based on Deblurring and Deep Learning Techniques

Detecting snow-covered solar panels is crucial as it allows us to remove snow using heating techniques more efficiently and restores the photovoltaic system to proper operation. This paper presents classification and detection performance analyses for snow-covered solar panel images. The classification analysis consists of two cases, and the detection analysis consists of one case based on three backbones. In this study, five deep learning models, namely visual geometry group-16 (VGG-16), VGG-19, residual neural network-18 (RESNET-18), RESNET-50, and RESNET-101, are used to classify solar panel images. The models are trained, validated, and tested under different conditions. The first case of classification is performed on the original dataset without preprocessing. In the second case, extreme climate conditions are simulated by generating motion noise; furthermore, the dataset is replicated using the upsampling technique to handle the unbalancing issue. For the detection case, a region-based convolutional neural network (RCNN) detector is used to detect the three categories of solar panels, which are all_snow, no_snow, and partial. The dataset of these categories is taken from the second case in the classification approach. Finally, we proposed a blind image deblurring algorithm (BIDA) that can be a preprocessing step before the CNN (BIDA-CNN) model. The accuracy of the models was compared and verified; the accuracy results show that the proposed CNN-based blind image deblurring algorithm (BIDA-CNN) outperformed other models evaluated in this study.

Detector Design and Performance Analysis for Target Detection in Subspace Interference

It is often difficult to obtain sufficient training data for adaptive signal detection, which is required to calculate the unknown noise covariance matrix. Additionally, interference is frequently present, which complicates the detecting issue. We provide a two-step method, termed interference cancellation before detection (ICBD), to address the issue of signal detection in the unknown Gaussian noise and subspace interference. The first involves projecting the test and training data to the interference-orthogonal subspace in order to suppress the interference. Utilizing traditional adaptive detector design ideas is the next stage. Due to the smaller dimension of the projected data, the ICBD-based detectors can function with little training data. The ICBD has two additional benefits over traditional detectors. Lower computational burden and proper operation with interference being in the training data are two additional benefits of ICBD-based detectors over conventional ones. We also give the statistical properties of the ICBD-based detectors and demonstrate their equivalence with the traditional ones in the special case of a large amount of training data containing no interference.