Low False Rate Research Articles

The characteristics of distal tears affect false lumen thrombosis rate after thoracic endovascular aortic repair for acute type B dissection.

A low false lumen thrombosis rate (FLTR) is common in patients with type B aortic dissection after they have undergone thoracic endovascular aortic repair, which indicates a poor long-term prognosis. This study aimed to establish a quantitative linear regression model to predict false lumen (FL) thrombosis accurately using morphological parameters. In this retrospective study, we included 59 patients diagnosed with acute type B aortic dissection between 2014 and 2017. Morphological parameters were measured. Univariable and multivariable linear regression analyses were performed, and a linear regression model relating FLTR with the number of re-entry tears was proposed. Ten patients were further chosen to validate the linear relationship, and idealized aortic dissection models were adopted for haemodynamic analysis. Only the total area and number of re-entry tears were negatively correlated with FL thrombosis (P < 0.001). Moreover, based on the univariable regression, the number of re-entry tears played a more crucial role in FLTR (R2 = 0.509 vs R2 = 0.298), and the linear relationship model was created as follows: thrombosis rate (%) = -11.25 × distal tear number + 105.24. This model was perfectly matched in 10 patients (concordance correlation coefficient = 0.880, P = 0.947). Moreover, when the total area of re-entry tears was constant, the net blood flow increased rapidly with an increase in the tear count. The number of re-entry tears could be a crucial related factor of FL thrombosis; the larger the number of re-entry tears is, the lower the possibility of thrombosis is.

Hybrid Statistical-Machine Learning for Real-Time Anomaly Detection in Industrial Cyber–Physical Systems

Critical industrial infrastructures are currently facing increasing cyberspace threats in their underlying information and communication systems. The advanced monitoring, control, and management functionalities of the industrial systems firmly rely on the reliable and secure operations of the industrial control system (ICS) network. This article characterizes the ICS network traffic and presents a scalable and efficient solution for real-time ICS network traffic anomaly detection, considering various forms of ICS anomaly events. The events due to the cyberattacks, malicious operating behaviors, and network anomalies can be effectively detected without sophisticated computational requirements and retrieval of communication protocols. The proposed hybrid statistical-machine learning model integrates a seasonal autoregressive integration moving average (SARIMA)-based dynamic threshold model and a long short-term memory (LSTM) model to jointly identify the abnormal traffic patterns with low false omission rates. The proposed solution is extensively evaluated at a realistic ICS cyber–physical system (CPS) testbed, and the numerical results confirm its high detection accuracy and low computational complexity. Note to Practitioners—This article was motivated by the challenge of real-time anomaly detection in industrial cyber–physical systems (CPSs). The existing industrial control system (ICS) network anomaly detection solutions are generally carried out based on a single model based on the historian database and cannot dynamically classify the abnormal conditions in a real-time fashion. A novel hybrid statistical-machine learning model is developed that integrates a seasonal autoregressive integration moving average (SARIMA)-based dynamic threshold model and a long short-term memory (LSTM) model to jointly identify the anomalous events through traffic pattern analysis. The proposed anomaly detection solution can efficiently provide accurate detection for cyberattacks, malicious operating behaviors, and network anomalies while meeting the real-time requirements of ICS networks. The proposed solution can be deployed in the realistic ICS CPSs, e.g., the power generation system, gas pipeline systems, and urban railway transportation systems. The preliminary numerical results obtained from the ICS-CPS testbed suggested that it can provide high detection accuracy with low computational complexity and, hence, can be adopted with minimal deployment hurdles.

Image-based surface scratch detection on architectural glass panels using deep learning approach

Abstract As a transparent and traditional building material, glass products such as glass facade are vital components of buildings. However, the surface scratches generated in the manufacturing process or emerging in the service stage such as windborne debris impacts may lead to remarkable strength degradation of glass material. In order to assess the fracture possibility of glass components, the size and number of scratches should be monitored during their lifecycle. Automatic scratch detection of architectural glass therefore remains a necessary task for civil engineers. A pixel-level instance segmentation model using Mask and region-based convolutional neural network (Mask R-CNN) was proposed for scratches detection on transparent glass surface. Images with scratches were firstly collected by a tailor-made automated microscopic camera scanning system to build the training and validation dataset. Test results demonstrate that the trained network is satisfactory, achieving a mean average precision of 96.5% with low missing and false rate under background interference. A comparison between the proposed model and another segmentation method YOLACT indicates that the proposed model has better performance in both detection and segmentation accuracy. The proposed deep learning-based approach can better support the development of non-contact defect assessment techniques for transparent building materials such as glass.

Efficient communication and EEG signal classification in wavelet domain for epilepsy patients

In this paper, we present an approach for the anticipation of electroencephalography (EEG) seizures using different families of wavelet transform. Different signal attributes are investigated to anticipate the seizure onset based on the wavelet transform. These attributes comprise amplitude, local mean, local median, local variance, derivative, and entropy of the wavelet-transformed signals. Different wavelet families are considered including Haar, Daubechies (db4, and db8), Symlets (Sym4), and Coiflets (Coif4) wavelets. The seizure prediction process is intended to be simple to be applied on a mobile application accompanying the patient to give him alerts of possible incoming seizures. The proposed approach is performed on long-term EEG recordings from the available CHB-MIT scalp dataset. It gives the best results in comparison with the other previous algorithms. It achieves a high sensitivity of 100% with Daubechies wavelet transform (db4) in addition to a low average False Prediction Rate (FPR) of 0.0818 h−1 and a high average Prediction Time (PT) of 38.1676 min. Therefore, it can help specialists for the prediction of epileptic seizures as early as possible.

Accurate detection of CNV based on single-nucleotide variants recalibration and image classification from whole genome sequencing

Copy number variations (CNVs) play an important role in the genome aberrations and human diseases. Comprehensive discovery of CNVs from whole genome sequencing data remains difficult because of low sensitivity and high false detective rate (FDR). We presented a novel framework which integrated SNV-based recalibration probabilistic model and image classification architecture (ImageCNV) for CNVs discovery. A Naive Bayesian model and a deep neural network InceptionV3 were adopted to infer candidate CNVs, and we utilize the benchmark datasets to evaluate the performance of our framework. ImageCNV yielded comparable sensitivity and lower FDR, complementing other methods based on different signals and providing a new perspective for the detection of CNVs. ImageCNV is freely available at https://github.com/minqing1/ImageCNV.

Machine Learning Techniques for Credit Card Fraud Detection

The term “fraud”, it always concerned about credit card fraud in our minds. And after the significant increase in the transactions of credit card, the fraud of credit card increased extremely in last years. So the fraud detection should include surveillance of the spending attitude for the person/customer to the determination, avoidance, and detection of unwanted behavior. Because the credit card is the most payment predominant way for the online and regular purchasing, the credit card fraud raises highly. The Fraud detection is not only concerned with capturing of the fraudulent practices, but also, discover it as fast as they can, because the fraud costs millions of dollar business loss and it is rising over time, and that affects greatly the worldwide economy. . In this paper we introduce 14 different techniques of how data mining techniques can be successfully combined to obtain a high fraud coverage with a high or low false rate, the Advantage and The Disadvantages of every technique, and The Data Sets used in the researches by researcher

Convolutional neural networks for automated microseismic detection in downhole distributed acoustic sensing data and comparison to a surface geophone array

ABSTRACTDistributed acoustic sensing is a growing technology that enables affordable downhole recording of strain wavefields from microseismic events with spatial sampling down to ∼1 m. Exploiting this high spatial information density motivates different detection approaches than typically used for downhole geophones. A new machine learning method using convolutional neural networks is described that operates on the full strain wavefield. The method is tested using data recorded in a horizontal observation well during hydraulic fracturing in the Eagle Ford Shale, Texas, and the results are compared to a surface geophone array that simultaneously recorded microseismic activity. The neural network was trained using synthetic microseismic events injected into real ambient noise, and it was applied to detect events in the remaining data. There were 535 detections found and no false positives. In general, the signal‐to‐noise ratio of events recorded by distributed acoustic sensing was lower than the surface array and 368 of 933 surface array events were found. Despite this, 167 new events were found in distributed acoustic sensing data that had no detected counterpart in the surface array. These differences can be attributed to the different detection threshold that depends on both magnitude and distance to the optical fibre. As distributed acoustic sensing data quality continues to improve, neural networks offer many advantages for automated, real‐time microseismic event detection, including low computational cost, minimal data pre‐processing, low false trigger rates and continuous performance improvement as more training data are acquired.

Transient stability assessment model with improved cost‐sensitive method based on the fault severity

For power system transient stability assessment (TSA), the consequences of misclassifications of unstable samples are more serious than those of stable samples. Therefore, the cost-sensitive method is introduced into TSA. A constant weight is set for all unstable training samples by the conventional cost-sensitive method. However, the fault severities of the training samples are different. Thus, the impacts of the samples with different fault severities on assessment rules are also different. Therefore, an improved cost-sensitive method based on the fault severity is proposed. Firstly, the fault severity of each sample is calculated by its fault clearing time. On this basis, different cost-sensitive coefficients are assigned to the samples with different fault severities. The weights of critical samples are larger than those of non-critical ones. Then, a stacked sparse auto-encoder model based on the improved cost-sensitive method is built through modifying the loss function. Finally, the proposed method is tested in an actual system. Compared with the traditional cost-sensitive method, the results show that the improved cost-sensitive method has better performance. Also, it has the advantages of the low false rate of unstable samples, high-assessment accuracy, and strong anti-noise ability.

RETRACTED ARTICLE: An optimized deep learning network model for EEG based seizure classification using synchronization and functional connectivity measures

Epilepsy is a brain disorder related to alteration in the nervous system which affects around 65 million people among the world’s population. Few works are focused on prediction of seizure relied on deep learning approaches, but the capability of optimal design has no longer been absolutely exploited. This work is focused on the seizure prediction obtained from long-short time records using optimized deep learning network model (ODLN). In this paper, the synchronization patterns and its feasibility of distinguishing the pre-ictal from inter-ictal states are examined by utilizing the interaction graph model as a functional connectivity measure. An optimized deep learning network with short- long-term memory is computed for the prediction of epileptic seizures occurrences. For, the modelling of ODLN, pre-analysis is performed with three modules and memory layers. It is finalized from these results; a two-layer ODLN is optimum to perform the epileptic seizure prediction for four different window sizes from 15 to 120 min. The assessment is implemented on the CHB-MIT Scalp EEG data set, providing 100% sensitivity and low false prediction rate ranges from 0.10 to 0.02 for seizure prediction. The proposed ODLN methodology reveals a notable increase in the performance rate of seizure prediction when compared with existing machine learning and Convolutional neural networks methods.

Design of network intrusion detection system based on parallel DPC clustering algorithm

With the advent of the era of big data, network intrusion detection systems based on K-means algorithm cannot meet the detection efficiency and detection speed requirements in big data environment. The DPC algorithm can be applied to high-dimensional network traffic and large-scale data application environments, but there are problems of large calculated amount and limited serial processing capability. Aiming at the problems of DPC algorithm, the DPC algorithm is adjusted firstly to improve the clustering accuracy of the algorithm. Then, the DPC algorithm a parallelised on the Spark platform, so that the processing ability and running speed of the DPC algorithm is greatly improved by running in parallel in the memory of multiple virtual machines. The experimental results show that the network intrusion detection system based on parallel DPC clustering algorithm has higher detection rate and lower false rate. The parallelisation clustering efficiency is much higher than the single-computer clustering efficiency.

Determination of optimal time for reading of rapid urease test diagnosis of Helicobacter pylori.

The study aimed to find the best reading time for the best accuracy of RUT in optimal time to obtain faster results with lower false rates and consequently save time in commencing treatment of peptic ulcers. Rapid urease test is well known to be an accurate test for H.pylori detection in tissue biopsies. Patients with GI problems referring to a university hospital in Tehran who underwent endoscopy and biopsy were entered in the project and three samples of mucosal tissue were captured from the lesser curvature, the antrum and the body of stomach. We found 39.6% sensitivity and 95% specificity for the named test in the first 5 minutes as well as PPV = 95.5% and NPV = 37.3% while the accuracy was 54.79%. Except for the specificity which was constantly 95% in all RUT reading times, other diagnostic parameters increased as time went on. The PPV was also higher than 97% after 10 minutes. The highest values of sensitivity, specificity, PPV, NPV and accuracy were achieved after 12 hours including 88.7%, 95%, 97.9%, 76% and 90.41%, respectively. To conclude, it seems that there are many different ideas with respect to the rapid urease test in H.pylori detection. However, the current study recommends reading the test optimally after 12 hours but it is suggested more multidisciplinary studies with bigger sample size be carried out to obtain better and more reliable results to be able to generalize in this regard.

Recovering value from single stream material recovery facilities – An outbound contamination analysis in Florida

The single stream recycling (SSR) program is a process in which all recyclable materials are deposited into a single collection bin. SSR has gained popularity in the U.S. due to its inherent abilities in waste collection, and specifically, in Florida, more than twenty counties have recently switched their recycling program from dual stream recycling (DSR) to SSR. Despite a more efficient collection process, mixing all recyclable materials into a single bin can lead to cross contamination even before reaching material recovery facilities (MRFs). This study aims to provide a better understanding of the sorting process and equipment in MRFs, and the impact of the SSR program on contamination rates in outbound materials that were processed through Florida’s recycling systems. First, we investigate the audit data obtained from a currently operating MRF in Florida using mass flow analysis to identify the most problematic recyclable streams and the processes with low efficiency and high false separation rates. According to our results, the sorting rates of mixed paper, glass and plastics are under the industry standards. Moreover, we investigate the outbound contamination rates of 35 old corrugated cardboard (OCC) and 266 old newsprints (ONP) samples obtained from four currently operating MRFs in Florida. Based on the results, only 31.4% of OCC samples and none of the ONP samples were within the accepted mills’ standards for contamination rates. This study provides valuable insights for lowering contamination and raising the end-product quality by identifying the problematic contaminants and processes in sorting and separation in MRFs.

Validation of a Biochip Chemiluminescent Immunoassay for Multi-Mycotoxins Screening in Maize (Zea mays L.)

A biochip chemiluminescent immunoassay was validated for multi-mycotoxins screening in maize. Screened mycotoxins were aflatoxins B1 (AFB1) and G1 (AFG1) ochratoxin A (OTA), zearalenone (ZEA), toxin T2 (T2), fumonisins (sum of FB1 and FB2) and deoxynivalenol (DON). The method included a single extraction step with acetonitrile:methanol:water (50:40:10, v/v/v). Spiked samples were fortified at 250 μg/kg for FB1 + FB2, 1 μg/kg for AFB1 and AFG1, 1.5 μg/kg for OTA, 50 μg/kg for ZEA, 25 μg/kg for T2 and 375 μg/kg for DON. The chemiluminescent signal of discrete test regions on the biochip is expressed in Relative Light Unit, and this value differs according to the mycotoxins’ concentration. Threshold value and the cut-off were calculated. Low false results rate was achieved (< 5%) and the obtained precision data is in agreement with EU legislation performance criteria. This assay revealed to be a valuable and cost-effective screening method for simultaneous semi-quantification of mycotoxins.

Early prediction of epileptic seizures using a long-term recurrent convolutional network

BackgroundA seizure prediction system can detect seizures prior to their occurrence and allow clinicians to provide timely treatment for patients with epilepsy. Research on seizure prediction has progressed from signal processing analyses to machine learning. However, most prediction methods are hand-engineered and have high computational complexity, increasing the difficulty of obtaining real-time predictions. Some forecasting and early warning methods have achieved good results in the short term but have low applicability in practical situations over the long term. New methodsFirst, electroencephalogram (EEG) time series were converted into two-dimensional images for multichannel fusion. A feasible method, a long-term recurrent convolutional network (LRCN), was proposed to create a spatiotemporal deep learning model for predicting epileptic seizures. The convolutional network block was used to automatically extract deep features from the data. The long short-term memory (LSTM) block was incorporated into learning a time sequence for identifying the preictal segments. New network settings and a postprocessing strategy were proposed in the seizure prediction model. ResultsThe deep seizure prediction model achieved an accuracy of 93.40%, prediction sensitivity of 91.88% and specificity of 86.13% in segment-based evaluations. For the event-based evaluations, 164 seizures were predicted. The proposed method provides high sensitivity and a low false prediction rate (FPR) of 0.04 F P/h. Comparison with existing methodsWe employed different methods, including the LRCN, deep learning and traditional machine learning methods, and compared them using the same data in this paper. Overall, the LRCN offers approximately 5–9% increased sensitivity and specificity. ConclusionThis study describes the LRCN network for analyzing EEG data to predict epileptic seizures, thereby enabling the implementation of an early warning system that detects epileptic seizures before they occur in clinical applications.

GLASS3: A Standalone Multiscale Seismic Detection Associator

Abstract The automated global real‐time association of phase picks into seismic sources comes with unique challenges when simultaneously monitoring at local, regional, and global scales. High spatial variability in seismic station density, transitory seismic data availability, and time‐varying noise characteristics of individual stations must be considered in the design of an associator that is fast and accurate with a low false association rate. These challenges are particularly apparent at the U.S. Geological Survey National Earthquake Information Center (NEIC), which monitors seismicity in near‐real time on local, regional, and global scales using seismic data from roughly 2100 real‐time seismic stations. To fully leverage this large dataset, NEIC developed a standalone self‐configuring seismic phase associator, GLobal ASSociator 3 (GLASS3) that simultaneously processes variably scaled 3D association webs, each with a unique set of nucleation criteria (e.g., nucleation stack threshold). GLASS3 has many useful features for real‐time monitoring including its computational efficiency, instantaneous pick processing, and on‐the‐fly configurability such as the creation and removal of targeted association webs and updates to supporting station metadata. GLASS3 runs both as part of a real‐time event processing system and as a configurable standalone associator that can be applied to a large variety of seismic problems. Here, we describe the GLASS3 algorithm and demonstrate (including input data and configuration files) its use in associating phase‐ambiguous picks on multiple scales.

Full Sibling Testing Based on NGS-SNP Genotyping Method and IBS Strategy.

Objective To evaluate the effectiveness of single nucleotide polymorphism （SNP） genoty-ping in combination with identity by state （IBS） strategy in full sibling testing. Methods Thirty-five blood samples were collected from a four-generation family. Ninety autosomal SNPs were genotyped using Precision ID Identity Panel. The distribution of IBS scores for full siblings and other relationships were calculated and compared. The relationships were determined using Fisher discriminant function and threshold method, respectively. Results Based on family members and previous research, 44, 30, 111, 71 and 1 000 pairs of full siblings （FS）, grandparent-grandchild （GG）, uncle/aunt-nephew/niece （UN）, first cousins （FC） and unrelated individuals （UI） were obtained, respectively. The average IBS scores were 148, 130, 132, 124 and 120, respectively. Except for the GG and UN pairs, the distribution differences among the other relationships had statistical significance （P<0.05）. The false rates of Fisher discriminant function to determine relationships were 1.3%, 22.3%, 17.0% and 38.7% for FS, GG, UN and FC, respectively. Based on the simulation data, the thresholds t1=128 and t2=141 were recommended to determine full sibling relationships （the false rate ≤0.05%）. Conclusion The 90 SNP genetic markers included in the Precision ID Identity Panel meet the testing requirements for full sibling relationships. The threshold method based on IBS has a relatively lower false rate and is more flexible.

The use of artificial neural network analysis can improve the risk-stratification of patients presenting with suspected deep vein thrombosis.

Artificial neural networks are machine-learning algorithms designed to analyse data without a pre-existing hypothesis as to any associations that may exist. This technique has not previously been applied to the risk stratification of patients referred with suspected deep vein thrombosis (DVT). Current assessment is usually with a points-based clinical score, which may be combined with a D-dimer blood test. A neural network was trained to risk-stratify patients presenting with suspected DVT and its performance compared with existing tools. Data from 11490 cases of suspected DVT presenting consecutively between 1 January 2011 and 31 December 2017 were analysed, and 7080 for whom all components of the Wells' score, a D-dimer and an ultrasound result were available were included in the analysis. The data were broken into a training set of 5270 patients, used to develop the algorithm, and a testing set of 1810 patients to assess performance of the trained algorithm. This network was able to exclude DVT without the need for ultrasound in significantly more patients than existing risk assessment scores, whilst retaining very low false negatives rates. More generally, this approach may improve the analysis of complex data to support decision-making in other areas of clinical medicine.

Automatic lung nodule detection using multi-scale dot nodule-enhancement filter and weighted support vector machines in chest computed tomography.

A novel CAD scheme for automated lung nodule detection is proposed to assist radiologists with the detection of lung cancer on CT scans. The proposed scheme is composed of four major steps: (1) lung volume segmentation, (2) nodule candidate extraction and grouping, (3) false positives reduction for the non-vessel tree group, and (4) classification for the vessel tree group. Lung segmentation is performed first. Then, 3D labeling technology is used to divide nodule candidates into two groups. For the non-vessel tree group, nodule candidates are classified as true nodules at the false positive reduction stage if the candidates survive the rule-based classifier and are not screened out by the dot filter. For the vessel tree group, nodule candidates are extracted using dot filter. Next, RSFS feature selection is used to select the most discriminating features for classification. Finally, WSVM with an undersampling approach is adopted to discriminate true nodules from vessel bifurcations in vessel tree group. The proposed method was evaluated on 154 thin-slice scans with 204 nodules in the LIDC database. The performance of the proposed CAD scheme yielded a high sensitivity (87.81%) while maintaining a low false rate (1.057 FPs/scan). The experimental results indicate the performance of our method may be better than the existing methods.

Design and performance analysis of various feature selection methods for anomaly‐based techniques in intrusion detection system

Intrusion detection system (IDS) is essential for the network; the intruder can steal sensitive information about networks. The IDS must have the ability to take care of large and real‐time data. The predicted rate must be high based on the available attribute. This work deals with a real intrusion detection problem, by its behavior. In this paper, we developed a hybrid model, which can detect intrusion by its action. We used an NSL‐KDD data set, the multiclass problem and binary problems are 25% tested. This model can be used to guess the availability of intrusion, able to determine the scope of intrusions based on the transaction of data in the network; training requires optimal features of a network transaction. The accuracy of the model is better for both binary class for the multiclass in NSL‐KDD data set. The complication of false data alarm rates is the most significant challenge in the IDS system, and it may be the low false rate or high false rate. Proposed work also addresses this problem. The first step that data will be filtered by Vote algorithm, the Information Gain will get associated with a base learner, to choose the necessary features, which directly affects the accuracy of the model. It uses the following classifier: RandomTree, REPTree, RandomForrest AdaBoostM1, Meta Pagging, DesicionStump, J48, LMT, Bagging, and Naive Bayes. On the based on the proposed model, it is observed as low false rate, high accuracy.

Deep Learning Based Automated Attendance System

Abstract A significant portion of the time allocated to a faculty for teaching purposes is consumed on the task of taking attendance of the students. This is an issue because it takes the valuable time of teachers which could be spent on more productive tasks such as teaching and interacting with students. In excess to the increase in chaos and loss of decorum in the classroom environment, the presence of proxy attendance also plagues the existing method of manual attendance keeping. To counter these issues, this paper proposed the Deep Learning Assisted Attendance System (DPAAS); which keeps track of students attending a particular class with the help of a continuous stream of pictures captured from a video streaming device located inside a classroom connected to the remote server. The proposed DPAAS method reduces the amount of time spent by the faculty on taking attendance, and leads to a reduction in chaos inside a classroom. DPASS is proposed handles the issues in existing systems such as multi-class identification for multiple individuals in a classroom, occlusion and differing light scenarios. The DPAAS methodology compares the results of the state of art algorithms, and uses the best fit architecture which provides the lowest false rate on evaluation. There is no need of user interaction in the proposed DPAAS. Experimental results show that the proposed DPAAS method gives 94.66% accuracy which is better than the other existing methods.