Low Test Accuracy Research Articles

From one-dimensional to multidimensional map neural networks

In the contemporary landscape, artificial neural networks transcend their traditional role of function approximation and have found diverse applications in fields such as image classification, machine translation, speech recognition, and natural language processing. In some datasets, traditional architectures exhibit low test and training accuracy, coupled with high loss and prolonged training times. This study aims to introduce innovative neural network architectures that outperform conventional models. The research presents a novel framework integrating one-dimensional and multidimensional map neural networks, consisting of three distinct architectures: 1D-Map, 2D-Map, and 3D-Map. A systematic performance comparison with traditional models is conducted following the implementation of these architectures. The evaluation spans four datasets, encompassing the domains of heat treatment of electroless Ni-P nano coatings, letter recognition, combined cycle power plants, and Seoul bike sharing demand. In the heat treatment dataset, the proposed 3D-Map architecture reached 0.055 higher average test accuracy than traditional MLP architecture. In the letter recognition dataset, the test accuracy of 3D-Map architecture was 0.0523 higher than the test accuracy of the LSTM architecture. In the combined cycle power plant dataset, the test accuracy of the 3D-Map architecture was 0.0612 more than the test accuracy of the MLP architecture. In the Seoul bike-sharing demand dataset, the test accuracy of the 2D-Map architecture was 0.0696 higher than the test accuracy of the LSTM architecture. The study's findings underscore the consistently superior performance of the proposed architectures compared to their traditional counterparts.

Enhancing intrusion detection: a hybrid machine and deep learning approach

The volume of data transferred across communication infrastructures has recently increased due to technological advancements in cloud computing, the Internet of Things (IoT), and automobile networks. The network systems transmit diverse and heterogeneous data in dispersed environments as communication technology develops. The communications using these networks and daily interactions depend on network security systems to provide secure and reliable information. On the other hand, attackers have increased their efforts to render systems on networks susceptible. An efficient intrusion detection system is essential since technological advancements embark on new kinds of attacks and security limitations. This paper implements a hybrid model for Intrusion Detection (ID) with Machine Learning (ML) and Deep Learning (DL) techniques to tackle these limitations. The proposed model makes use of Extreme Gradient Boosting (XGBoost) and convolutional neural networks (CNN) for feature extraction and then combines each of these with long short-term memory networks (LSTM) for classification. Four benchmark datasets CIC IDS 2017, UNSW NB15, NSL KDD, and WSN DS were used to train the model for binary and multi-class classification. With the increase in feature dimensions, current intrusion detection systems have trouble identifying new threats due to low test accuracy scores. To narrow down each dataset’s feature space, XGBoost, and CNN feature selection algorithms are used in this work for each separate model. The experimental findings demonstrate a high detection rate and good accuracy with a relatively low False Acceptance Rate (FAR) to prove the usefulness of the proposed hybrid model.

Enhanced selective discrimination of point-mutated viral RNA through false amplification regulatory direct insertion in rolling circle amplification

Coronaviruses are single-stranded RNA viruses with high mutation rates. Although a diagnostic method for coronaviruses has been developed, variants appear rapidly. Low test accuracy owing to single-point mutations is one of the main factors in the failure to prevent the early spread of coronavirus infection. Although reverse transcription-quantitative polymerase chain reaction can detect coronavirus infection, it cannot exclude the possibility of false positives, and an additional multiplexing kit is needed to discriminate single nucleotide polymorphism (SNP) variants. Therefore, in this study, we introduced a new nucleic acid amplification method to determine whether an infected person has a SNP mutation using a lateral flow assay (LFA) as a point-of-care test. Unlike traditional DNA amplification methods, direct insertion into rolling circle amplification amplifies the target genes without false amplification. After SNP-selective nucleic acid amplification, nuclease enzymes are used to make double-stranded DNA fragments that the LFA can detect, where specific mismatched DNA is found and cleaved to show different signals when a SNP-type is present. Therefore, wild- and SNP-type variants can be selectively detected. In this study, the limit of detection was 400 aM for viral RNA, and we successfully identified a dominant SNP variant selectively. Clinical tests were also conducted.

A deep learning technique for intrusion detection system using a Recurrent Neural Networks based framework

In recent years, the spike in the amount of information transmitted through communication infrastructures has increased due to the advances in technologies such as cloud computing, vehicular networks systems, the Internet of Things (IoT), etc. As a result, attackers have multiplied their efforts for the purpose of rendering network systems vulnerable. Therefore, it is of utmost importance to improve the security of those network systems. In this study, an IDS framework using Machine Learning (ML) techniques is implemented. This framework uses different types of Recurrent Neural Networks (RNNs), namely, Long-Short Term Memory (LSTM), Gated Recurrent Unit (GRU) and Simple RNN. To assess the performance of the proposed IDS framework, the NSL-KDD and the UNSW-NB15 benchmark datasets are considered. Moreover, existing IDSs suffer from low test accuracy scores in detecting new attacks as the feature dimension grows. In this study, an XGBoost-based feature selection algorithm was implemented to reduce the feature space of each dataset. Following that process, 17 and 22 relevant attributes were picked from the UNSW-NB15 and NSL-KDD, respectively. The accuracy obtained through the test subsets was used as the main performance metric in conjunction with the F1-Score, the validation accuracy, and the training time (in seconds). The results showed that for the binary classification tasks using the NSL-KDD, the XGBoost-LSTM achieved the best performance with a test accuracy (TAC) of 88.13%, a validation accuracy (VAC) of 99.49% and a training time of 225.46 s. For the UNSW-NB15, the XGBoost-Simple-RNN was the most efficient model with a TAC of 87.07%. For the multiclass classification scheme, the XGBoost-LSTM achieved a TAC of 86.93% over the NSL-KDD and the XGBoost-GRU obtained a TAC of 78.40% over the UNSW-NB15 dataset. These results demonstrated that our proposed IDS framework performed optimally in comparison to existing methods.

Application of cyclodextrin/polyethylene glycol in blood-storage tools for liquid biopsy

• Inhibition of wild-type DNA release from leukocytes for liquid biopsy. • γ-Cyclodextrin (γ-CyD)/polyethylene glycol (PEG) is a potential additive. • γ-CyD/PEG suppress the wild-type DNA release from leukocytes. • γ-CyD/PEG is applicable for blood collection tubes for liquid biopsy. In recent years, liquid biopsy has garnered considerable attention as a valuable diagnostic technology for cancer. An analysis target of liquid biopsy is cell-free DNA; however, contamination with wild-type DNA released from injured leukocytes leads to low test accuracy. Thus, there is a need for blood-storage tools that inhibit the release of wild-type DNA from leukocytes. In the present study, we demonstrated that γ-cyclodextrin (γ-CyD) and polyethylene glycol (PEG) synergistically inhibit DNA release from cells by inhibiting cell aggregation. In addition, formation of a network structure of PEG via the inclusion complexation of γ-CyD with two PEG chains is probably important for the inhibitory effects on DNA release. These findings indicate that the combination of γ-CyD and PEG has potential as a blood-storage tool for liquid biopsy.

Target Detection and Network Optimization: Deep Learning in Face Expression Feature Recognition

Facial expression recognition technology has been more and more in demand in security, entertainment, education, medical, and other domains as artificial intelligence has advanced, and face expression recognition technology based on deep learning has become one of the research hotspots. However, there are still some issues with the existing deep learning convolutional neural network; the feature extraction technique has to be improved, and the design of the detailed network structure needs to be optimized. It is critical to do more research on the deep learning convolutional neural network model in order to increase the accuracy of face facial expression detection. In this paper, a deep learning convolutional neural network structure combining VGG16 convolutional neural network and long and short-term memory networks is designed to address the shortcomings of existing deep learning methods in face expression recognition, which are prone to overfitting and gradient disappearance, resulting in low test accuracy. This structure easily and effectively collects facial expression information and then classifies the retrieved features using a support vector machine to detect face expressions. Finally, the fer2013 dataset is used to train face expression recognition, and the results demonstrate that the built deep convolutional neural network model can effectively increase face expression identification accuracy.

Utilization of point-of-care tests among general practitioners, a cross-sectional study

BackgroundPoint-of-care testing (POCT) has numerous potential benefits to improve health care service, especially in resource-limited settings. We aim to identify which POC-tests (POCTs) of laboratory parameters are known, employed, and rated as useful by general practitioners (GPs).MethodsA questionnaire with 27 POCTs was posted to a random selection of GPs (n = 451) in Saxony, Germany.ResultsA total of 208 GPs replied (response rate 46.1%). Out of 27 POCTs, each GP knew an average of 20.3 as laboratory parameters and 9.2 as POCTs. Urine test strips (99.0%), blood glucose test (98.1%), and Troponin I/T (86.4%) were the best-known, followed by INR/Quick (82.5%), Microalbumin (79.1%), and D-dimer (78.6%) POCTs. Yet, solely 0 to 13 POC tests were actually used (mean value 4.6). Urine test strips were employed most frequently (97.6%), followed by blood glucose test (94.7%), Troponin I/T (57.8%), Microalbumin (57.3%), and INR/Quick POCTs (41.7%). Heart fatty binding protein (H-FABP), Syphilis, Coeliac disease, and Malaria appeared as the least frequently used POCTs. The majority of the GPs declared 14 of the 27 POCTs to be useful.Discussion/conclusionThe most recurrently employed POCTs are those for diagnosing or monitoring diabetes mellitus, ensued by POCTs addressing acute cardiovascular diseases (Troponin I/T, D-dimer) or monitoring the therapy of infectious diseases or the anticoagulant therapy. POCTs most often rated as useful by GPs are also widely known and frequently used. Nonetheless, the majority of GPs rate only a very limited number of POCTs as useful. Frequent concerns might be low economic benefit, over-reliance, and test accuracy coming along with the complex implementation of the tests requiring technical skills, accurate storage, and the correct interpretation of test results.Trial registrationIn accordance with the (Model) Professional Code for Physicians in Germany, neither human body materials nor data that can be assigned to a specific human being are used in our study. A declaration of no objection from the Ethics Committee of the Martin-Luther University Halle-Wittenberg (Medical Faculty) confirms no professional or ethical concerns due to completely anonymized data collection and analysis. Our study was therefore not registered in a corresponding registry.

Network with Sub-networks: Layer-wise Detachable Neural Network

In this paper, we introduce a network with sub-networks: a neural network whose layers can be detached into sub-neural networks during the inference phase. To develop trainable parameters that can be inserted into both base-model and sub-models, first, the parameters of sub-models are duplicated in the base-model. Each model is separately forward-propagated, and all models are grouped into pairs. Gradients from selected pairs of networks are averaged and used to update both networks. With the Modified National Institute of Standards and Technology (MNIST) and Fashion-MNIST datasets, our base-model achieves identical test-accuracy to that of regularly trained models. However, the sub-models result in lower test-accuracy. Nevertheless, the sub-models serve as alternative approaches with fewer parameters than those of regular models.

Paleoclimatic indication of X-ray fluorescence core-scanned Rb/Sr ratios: A case study in the Zoige Basin in the eastern Tibetan Plateau

The X-ray fluorescence (XRF) core-scanned Rb/Sr ratios of lake sediments have been widely used as a proxy for chemical weathering intensity and past climate change. However, some factors could affect Rb/Sr ratios, causing misinterpretation and limiting its application. In this study, we present a high-resolution XRF core-scanned Rb/Sr record of core ZB13-C1 from the Zoige Basin in the eastern Tibetan Plateau. To validate its application, we correlated this record with the chemical index of alteration (CIA) and other paleoclimatic proxies. Our results showed that (1) the core-scanned Rb/Sr ratios and CIA were reliable proxies of chemical weathering intensity in fine-grained sedimentary sequences; (2) the low values of core-scanned Rb/Sr ratios and CIA were significantly correlated with high total organic carbon content, arboreal pollen content, carbonate content, and C/N ratios, confirming its reliability as a proxy for the Asian summer monsoon intensity; (3) the core-scanned Rb/Sr ratios at core depths of 25–0.3 and 56–17 m were unable to reliably reflect chemical weathering intensity due to both the grain-size effect and the low test accuracy. Our study highlights the need for mutual verification of multiple indicators before accurately applying Rb/Sr as a paleoclimatic proxy in other similar study areas.

Evidence-Based Diagnostic Test Accuracy of History, Physical Examination, and Imaging for Intussusception: A Systematic Review and Meta-analysis.

Intussusception is the most common cause of pediatric small bowel obstruction. Timely and accurate diagnosis may reduce the risk of bowel ischemia. We quantified the diagnostic test accuracy of history, physical examination, abdominal radiographs, and point-of-care ultrasound. We conducted a systematic review for diagnostic test accuracy of history, physical examination, and imaging concerning for intussusception. Our literature search was completed in June 2019. Databases included Medline via Ovid, Embase, Scopus, and Wiley Cochrane Library. We conducted a second review of the literature up to June 2019 for any additional studies. Inclusion criteria were younger than 18 years and presenting to the emergency department for abdominal complaints, consistent with intussusception. We performed data analysis using mada, version 0.5.8. We conducted univariate and bivariate analysis (random effects model) with DerSimonian-Laird and Reitsma model, respectively. QUADAS-2 was used for bias assessment. The literature search identified 2639 articles, of which 13 primary studies met our inclusion criteria. Abdominal pain, vomiting, and bloody stools had positive likelihood ratios LR(+) between 1 and 2, whereas the negative likelihood ratio, LR(-), ranged between 0.4 and 0.8. Abnormal abdominal radiograph had LR(+) of 2.5 and LR(-) of 0.20, whereas its diagnostic odds ratio was 13. Lastly, point-of-care ultrasound had LR(+) of 19.7 and LR(-) of 0.10. The diagnostic odds ratio was 213. History and physical examination had low diagnostic test accuracy. Abdominal radiographs had low diagnostic test accuracy, despite moderate discriminatory characteristics. Point-of-care ultrasound had the highest diagnostic test accuracy to rule in or rule out intussusception.

An Optimization Analysis on Testing Method for Cooking Oil Fumes Control Equipment in China

Cooking Oil fumes have become one of the most important pollutants in China's society. Oil fume purification equipment has emerged as the times require. China has issued inspection standards for the quality control of such products, but these standards have some drawbacks such as poor operation, low test accuracy, and poor data reproducibility. According to actual work experience, this method is based on the sampling process and the extraction solution. The recommendations for optimization of the air-tightness, air-tightness, exhaust emissions, and testing instruments are provided in 5 areas, which provide references for the quality inspection of such products.

Factors affecting cell-free DNA fetal fraction and the consequences for test accuracy

Introduction: Biological factors are known to influence the fetal fraction (FF) of cell-free DNA and may also influence the accuracy of non-invasive prenatal testing.Material and methods: NIPT from 5267 mixed risk women across three specialist clinics in Australia were analyzed. Multivariable regression analysis was used to determine whether maternal characteristics, ultrasound, and placental biomarkers affect FF and test accuracy.Results: FF ranged from 4% to 37% (mean 11.6%). Body mass index (BMI), gestation, and placental biomarkers were found to be significant factors associated with FF. For each unit increase in BMI, the logarithmically transformed FF, (lnFF), mean value decreased by 0.027. Each week increases in gestation, lnFF increased by 0.023. Each unit increase in free BhCG, PAPPA, and PlGF, the lnFF increased by 0.065, 0.050, and 0.17, respectively. There was no significant association between FF with either maternal age or nuchal translucency. The false-positive cases and one false-negative case did not have lower FF than the true-positive cases.Discussion: The fetal fraction in maternal plasma cfDNA increased with gestational age, serum pregnancy-associated plasma protein A (PAPP-A), β-hCG, and PlGF and decreased with increasing maternal BMI. There was no significant correlation between low FF and test accuracy, when FF was above 4%.

Mexican Hat Wavelet Kernel ELM for Multiclass Classification.

Kernel extreme learning machine (KELM) is a novel feedforward neural network, which is widely used in classification problems. To some extent, it solves the existing problems of the invalid nodes and the large computational complexity in ELM. However, the traditional KELM classifier usually has a low test accuracy when it faces multiclass classification problems. In order to solve the above problem, a new classifier, Mexican Hat wavelet KELM classifier, is proposed in this paper. The proposed classifier successfully improves the training accuracy and reduces the training time in the multiclass classification problems. Moreover, the validity of the Mexican Hat wavelet as a kernel function of ELM is rigorously proved. Experimental results on different data sets show that the performance of the proposed classifier is significantly superior to the compared classifiers.

The role of rapid diagnostics in managing Ebola epidemics.

Ebola emerged in West Africa around December 2013 and swept through Guinea, Sierra Leone and Liberia, giving rise to 27,748 confirmed, probable and suspected cases reported by 29 July 2015. Case diagnoses during the epidemic have relied on polymerase chain reaction-based tests. Owing to limited laboratory capacity and local transport infrastructure, the delays from sample collection to test results being available have often been 2 days or more. Point-of-care rapid diagnostic tests offer the potential to substantially reduce these delays. We review Ebola rapid diagnostic tests approved by the World Health Organization and those currently in development. Such rapid diagnostic tests could allow early triaging of patients, thereby reducing the potential for nosocomial transmission. In addition, despite the lower test accuracy, rapid diagnostic test-based diagnosis may be beneficial in some contexts because of the reduced time spent by uninfected individuals in health-care settings where they may be at increased risk of infection; this also frees up hospital beds. We use mathematical modelling to explore the potential benefits of diagnostic testing strategies involving rapid diagnostic tests alone and in combination with polymerase chain reaction testing. Our analysis indicates that the use of rapid diagnostic tests with sensitivity and specificity comparable with those currently under development always enhances control, whether evaluated at a health-care-unit or population level. If such tests had been available throughout the recent epidemic, we estimate, for Sierra Leone, that their use in combination with confirmatory polymerase chain-reaction testing might have reduced the scale of the epidemic by over a third.

A cross sectional study evaluating screening using maternal anthropometric measurements for outcomes of childbirth in Ugandan mothers at term.

BackgroundBirth related newborn and maternal mortality/morbidity remains high in most of sub-Saharan Africa compared to the rest of the world. In this low income region there is a need for valid, low cost, easy to use mass screening tests. This study looked at the screening value of maternal: height, weight and pelvis height, for assessing the outcomes of parturition in Ugandan mothers at term.MethodsThis was a multi site cross-sectional study on mothers with singleton pregnancies in labour at various hospitals in different parts of Uganda. A summary of the details of the pregnancy, maternal height, weight and the delivery record were captured and analysed to generate descriptive and inferential (multilevel logistic regression analysis) and diagnostic (Receiver Operator Curve analysis) statistics.ResultsWe recruited 1146 mothers from all the study sites during the study period of whom 987 (86.13%) had normal deliveries and healthy babies. Mothers with adverse outcomes included 107 mothers that had caesarean section and 52 mothers who had vaginal deliveries with foetal Apgar score of ≤7 at 5 min of whom 11 had fresh still births. Maternal height (Adj OR 0.97, 95% CI 0.94–1.00) and maternal pelvis height (Adj OR 0.73, 95% CI 0.61–0.86) were significantly associated with adverse pregnancy outcomes. The combination of maternal: height (<150 cm), weight (>55.7 kg) and pelvis height (>8.95 cm) had the best diagnostic value with a combined area under the curve of 0.60.ConclusionsIt was observed that an increase in either maternal pelvis height or maternal height was associated with a significant reduction in adverse pregnancy outcomes. The cut off values of all three evaluated maternal anthropometric measurements were of low test accuracy as screening tests even when used together. Further research is needed to develop low cost screening tools for use in low income settings.

Psychometric properties of the DCD-Q-07 in children ages to 4–6

It is important to identify Developmental Coordination Disorder (DCD) early in a child's life to allow for proper and timely intervention and support, and to reduce the negative secondary consequences associated with this condition. In this study we assessed the psychometric properties (construct validity, concurrent validity, reliability and test accuracy) of the Developmental Coordination Disorder Questionnaire (DCD-Q-07) in preschool children. A community-based sample of children ages 4-6 (n=181) were screened for motor difficulties using the Movement Assessment Battery for Children (M-ABC-2). Use of the M-ABC-2 resulted in the identification of 29 children below the 15th percentile, which we classified as probable DCD. Parents of these children concurrently completed the DCD-Q-07 to report their child's motor performance. The DCD-Q-07 demonstrated high internal consistency for both the full scale (alpha=0.881) and each subscale: control during movement (alpha=0.813), fine motor and handwriting (alpha=0.869) and general coordination (alpha=0.728). Moderate correlations (r=0.47-0.63) were also seen between the subscales on the DCD-Q-07, the strongest correlation being between control during movement and general coordination (r=0.63). Based on published age and sex cut points, the DCD-Q-07 showed poor sensitivity (20.7%) but high specificity (92.1%) against the M-ABC-2. Overall agreement with the M-ABC-2 was low using ROC analysis (area under the curve=0.654). Although it is important to screen for DCD in young children, the DCD-Q-07 may not be the best choice as a screening tool for DCD in preschool children ages 4-6 due to its low test accuracy.

Capacitor-Coupled Built-Off Self-Test in Analog and Mixed-Signal Embedded Systems

Design-for-test (DfT) circuitry that employs differential terminals inherently suffers from an imbalance in the output of its differential pair. By providing the imbalanced differential test stimulus from the DfT circuitry, nonlinearity is eventually introduced in a differential mixed-signal circuit under test, resulting in low test accuracy and significant yield loss during production testing. Consequently, in only rare cases are attempts made to measure dynamic performance of differential mixed-signal circuits using a self-test approach. This brief suggests an efficient testing technique based on built-off self-test for differential analog and mixed-signal circuits. This technique precisely predicts individual device-under-test (DUT) specifications by resolving the imbalance problem using simple variable capacitors in loopback mode. The variable capacitor generates predefined imbalances to give different weights on the spectral loopback responses. Nonlinear models are derived to characterize the DUT specifications using the differently weighted loopback responses. The hardware measurement results can be used to validate that the proposed method should be able to replace the conventional test method.

Imbalance-Based Self-Test for High-Speed Mixed-Signal Embedded Systems

Precisely measuring specifications of differential analog and mixed-signal circuits is a difficult problem for self-test development because the imbalance introduced by the design-for-test circuitry on the differential signaling causes nonlinearity on the test stimulus, resulting in degrading device-under-test (DUT) performance. This problem triggers low test accuracy and serious yield loss. This brief proposes a novel test methodology to accurately predict individual DUT specifications by overcoming the imbalance problem with the imbalance generator, a radio-frequency transformer, and a programmable capacitor array based on a loopback test configuration. The imbalance generator produces spectral loopback responses of different weight. Nonlinear equations are then derived to characterize DUT specifications. Hardware measurement results show that this approach can be used to predict the specifications of a DUT in a production test.

Efficient Loopback Test for Aperture Jitter in Embedded Mixed-Signal Circuits

Accurate measurement of aperture jitter for high-speed data converters is a difficult problem, since aperture jitter should be precisely separated from other jitter components as well as additive noise. This problem results in low test accuracy and high-yield loss. This paper presents a novel methodology for accurately predicting aperture jitter using a cost-effective loopback methodology. By using an efficient spectral loopback scheme, aperture jitter is precisely separated from input jitter and clock jitter as well as additive noise present in the DUT. Hardware measurement results show that this approach can be effectively used to predict the aperture jitter of a DUT, with an 89% reduction in the prediction error compared with previous approaches.

Transformer-Coupled Loopback Test for Differential Mixed-Signal Dynamic Specifications

Loopback tests for a differential mixed-signal device under test (DUT) have rarely been attempted since any imbalance introduced by design-for-test (DfT) circuitry on differential signaling delivers an imperfect sinusoidal wave to the DUT input or output, thereby degrading the DUT performance. In addition, this methodology inherently suffers from fault masking. These problems trigger low test accuracy and serious yield loss. This paper presents a novel methodology for the efficient prediction of individual DUT dynamic performance parameters with a radio-frequency (RF) transformer in loopback mode to overcome the imbalance of DfT circuitry and the fault masking. Cascaded RF transformers with different multiplicative weights are selected in three combinations by a multiplexer to create three separate loopback responses. These responses are used to characterize the DUT dynamic performance. Hardware measurement results show that this approach can be effectively used to predict the specifications of a DUT.