Search

In the intensive care unit(ICU), excessive false alarms burden medical staff greatly, and cause a waste of medical resources as well. In order to alleviate false alarming in ICU, we constructed classification models based on convolutional neural network, which can directly deal with time series and avoid manually extracting features. Combining with grouping strategy, we tried two basic network structures, i.e. deep group convolutional neural network(DGCN) and embedded deep group convolutional network(EDGCN). After that, based on EDGCN, which was proved better, ensemble networks were constructed to elevate the performance further. Considering of the limited sample size, various data expansions were also tried. For comparison, we used a widely adopted index ‘Score’, which is defined as $100\times $ (TP+TN)/(TP+TN+FP $+ 5\times $ FN), as performance evaluation. Finally, we tested our model in the online sandbox, and got a score of 80.68. Although it is slightly lower than the best scores that have been reported, our models are end-to-end, which means the original time series can be automatically mapped into a binary output, without manual feature extraction. In addition, the innovative adoption of group convolution makes full use of information from multi-channel signals. Besides, this work provides ideas for relating network parameter selection with physiological or physical natures of signals. In the end, after further discussion, we also proposed potential elevation solutions.

Particle filter has been widely applied in nonlinear target tracking due to the ability to carry multiple hypothesis and relaxation of linearity/Gaussian assumption. In this paper, an adaptive double space-resampling particle filter is proposed to increase the efficiency and robustness of filtering by adjusting the sample size. The first resampling operation, adopted before the prediction of samples, generates a larger number of equal-weighted samples and some auxiliary samples to enhance the robustness of filtering. The second resampling, adopted between the prediction and updating step, decreases the sample size for weight updating which is the most time consumption part of particle filter. The particle space sampling technique is used in both space-resampling, which adjusts the sample size according to not only the weights of samples but also their spatial distribution. The efficiency of filtering is improved and the robustness of algorithm is enhanced, simultaneously. The degeneracy and sample impoverishment problems can be counteracted. Simulation and experiment contrast results demonstrate that the proposed method is robust and efficient.

Limited training data, high dimensionality, image complexity, and similarity between classes are challenges confronting hyperspectral image (HSI) classification often resulting in suboptimal classification performance. The capsule network (CapsNet) preserves the hierarchy between different parts of the entity in an image by replacing scalar representations with vectors and can address these aforementioned issues. Motivated by CapsNet, this article presents a novel end-to-end deep learning (DL) architecture, the hybrid capsule network (HCapsNet), for HSI classification. HCapsNet employs 2-D and 3-D convolutional neural networks (CNNs) to extract higher level spatial and spectral features. In order to establish a route between capsules in the lower layers to the most-related capsule in the higher layer, dynamic routing (DR) is used to identify several overlapped objects during training sessions. Hyperparameter optimization is performed using nested cross-validation (nested-CV) to ensure thorough generalization evaluation. The proposed HCapsNet significantly outperformed the state-of-the-art methods in terms of overall classification accuracy on three widely used hyperspectral datasets, Indian Pines dataset achieving ($>3\%$, $p< {1}\times 1\times 10^{-11}$), the University of Pavia dataset ($>4\%$, $p< {1}\times 1\times 10^{-9}$), the Salinas Valley dataset ($>3\%$, $p< {1}\times 1\times 10^{-10}$) when using only 1% of the data for training. The performance of all CNN-based approaches degraded significantly with smaller training sample sizes. HCapsNet, therefore, is demonstrated to offer significant advantages in HSI classification problems with low sample sizes.

In Monte Carlo based importance sampling estimations, Effective Sample Size (ESS) is an important index of simulation efficiency, since ESS can measure the divergence between the target distribution and the proposal distribution effectively, and thus is widely used to decide whether resampling is needed or not. Among several well-known variants of ESS, the Shannon entropy based perplexity has been widely used. In this paper, however, we propose a new ESS function (E-MIM) by using the message importance measure (MIM) instead of Shannon entropy. We show that E-MIM satisfies all of the five conditions for ESS generalizations. We also propose an MIM based divergence and investigate its approximation to E-MIM. Moreover, we present a resampling threshold selection method for the ratio between E-MIM and the corresponding actual sample size. Finally, we investigate the performance of E-MIM and other ESS functions through numerical simulations. By a particle filter experiment, we show that E-MIM outperforms other ESS functions in terms of mean-squared error.

Colored dissolved organic matter (CDOM) is an important biochemical state indicator of aquatic ecosystems. However, its retrieval from remote sensing datasets remains a challenging task due to its high spatiotemporal variability and interference from other water constituents. In this article, we aim to develop new CDOM inversion algorithms by taking advantage of a representative sparse learning algorithm known as least absolute shrinkage and selection operator (LASSO), which is applied to identify the optimal band arithmetic terms for the CDOM inversion and estimate the model parameters in a more global and robust manner than can statistics-based methods. Moreover, a two-stage inversion framework is presented to further enhance the stability of LASSO in addressing inadequate in situ sample circumstances. Within the framework, two different schemes are proposed to handle the band arithmetic terms information propagation between the two stages, for which two new algorithms are proposed. Experimental results obtained from the in situ bio-optical dataset and the synthesized dataset under various training sample sizes indicate that both of the proposed algorithms can deliver performance superior to six state-of-the-art CDOM inversion algorithms, and with less sensitivity to the training sample states. In addition, the results offered by the new algorithms also enjoy biochemical interpretability, revealing that the red/blue ratio terms are well suited for inverting the CDOM content on these datasets.

Online aerodynamic parameter estimation plays an important role in compensating control system of aircraft under parameter uncertainties and unknown disturbance. In this paper, stability and control derivatives of aircraft are estimated online via support vector regression-numerical differential(SVR-ND) method. Small-sample real-time flight data reflecting real-time aerodynamic characteristics of aircraft is processed as training samples. For the small-size training samples, SVR technique is used for aerodynamic modeling. To pursue good performance in both computation efficiency and estimation accuracy, offline parameter estimation simulations are performed to select training sample size. It is observed that parameter estimation accuracy is related to the number of training samples and the noise level of samples. After that, an empirical formula is proposed to select training sample size according to results of simulations. To adapt the variation of samples, empirical formulas to tune hyper-parameters of SVR are presented based on the estimation of noise variance of samples. Finally, aerodynamic parameters are obtained by numerical differential in real-time. In a simulated maneuver, the proposed method is applied to online aerodynamic parameter estimation, and a Monte Carlo simulation is carried out to validate the robustness of SVR-ND method. Results indicate that the proposed method could realize accurate and robust estimation of aerodynamic parameters online.

For most diseases, building large databases of labeled genetic data is an expensive and time-demanding task. To address this, we introduce genetic Generative Adversarial Networks (gGAN), a semi-supervised approach based on an innovative GAN architecture to create large synthetic genetic data sets starting with a small amount of labeled data and a large amount of unlabeled data. Our goal is to create a mechanism able to increase the sample size of the labeled data and generalize learning over different populations while keeping awareness of the quality of its own predictions. The proposed model achieved satisfactory results using real genetic data from different datasets and populations, in which the test populations may not have the same genetic profiles. The proposed model is self-aware and capable of determining whether a new genetic profile has enough compatibility with the data on which the network was trained and is thus suitable for prediction. The code and datasets used can be found at https://github.com/caio-davi/gGAN

Currently, statistical tests for random number generators (RNGs) are widely used in practice, and some of them are even included in information security standards. But despite the popularity of RNGs, consistent tests are known only for stationary ergodic deviations of randomness (a test is consistent if it detects any deviations from a given class when the sample size goes to infinity). However, the model of a stationary ergodic source is too narrow for some RNGs, in particular, for generators based on physical effects. In this article, we propose computable consistent tests for some classes of deviations more general than stationary ergodic and describe some general properties of statistical tests. The proposed approach and the resulting test are based on the ideas and methods of information theory.

In this paper, a thorough analysis of an optimal measured sample size and placement over the surface of a microstrip ring resonator (MRR) is conducted using a full-wave software and respective measurements using a vector network analyzer as well as a suitable analytical model to compute the unknown value of relative permittivity of a material under test. Using this systematic approach, we present our findings regarding the optimal sample-under-test size and placement over the surface of an MRR.

Sex estimation is used in forensic anthropology to assist in the identification of human remains. The estimation model, on the other hand, is usually one-of-a-kind and only applies to a specific group of people. In addition, the study on Asian population also limited. This paper proposed sex estimation model using Random Forest based on the length of bones in x-ray images of left hand taken from Asian dataset. All the nineteen bones of the hand were appraised in centimeter using free image software and the Random Forest model was performed using Matlab. The result shows that the Random Forest model produced the best average of accuracy percentage on age group 16 to 18 years old which is 91.67% compare to the other group. The result also reveals that only this age group can be used for sex estimation due to the results produced more than 80%. However, due to the limited sample size, this study may not precisely reflect the overall Asian population, which it may impact the data's interpretation.

This work improves upon a previously developed neural network modelling process that predicted waveguide parameters for the D-region ionosphere on two days [1]. The previous model was limited by manually determining the ideal set of transmitters (Tx) and receivers (Rx) and by computation time. An automatic quality assessment tool was developed to automatically evaluate the optimal network for each day [2]. We also obtained a 14x improvement in model training time by leveraging GPUs and improving the parallelization of the training process. These advancements allowed us to model 328 days across up to 21 paths. With this larger sample size, we show the model is capable of following expected seasonal trends. The model has also been adapted to be used with nighttime data, and is showing promising early results.

We propose a gradient-based simulation optimization approach for economic design of control charts. A generalized likelihood ratio method is applied to estimate the gradient. Two stochastic approximation algorithms with increasing sample size in iterations and randomized sample size are developed to determine an optimal upper control limit for exponentially weighted moving average control chart. Numerical results show that the proposed method is an effective approach for economic design of control charts.

Online lending platform offer benefits for underserved segment that need access to loans. The massive development on this industry because of the convenient access compared to bank or multi finance. OJK as the authorities expect this industry can support financial inclusion in Indonesia. This study designs to identify the Variables Affecting Consumers’ Acceptance of Online Lending Platform based on problems and opportunities. Variables related to opportunities are Familiarity and Interest Rate, while variables related to problems are Trust and Security. This study’s sample size determination method using estimation of a proportion, unit analysis is the individual customers, sampling technique using simple random sampling processed 145 respondents and processed the data using PLS-SEM. The results are Familiarity affects the acceptance of online lending platform. This research also produces recommendations for the development of the industry so it can support financial inclusion.

Telemedicine is of growing importance for the increasing number of patients with population aging, lack of medical resources in the countryside, and special situation such as the COVID-19 pandemic. Gait impairment is a major motor symptom for patients suffering from neurological disorders such as Parkinson’s disease (PD), and serves as an important indicator for early screening and diagnosis of the disease. Existing gait analysis methods typically require advanced equipment, trained professionals, and complex procedures. In this paper, we propose a method to classify the abnormal gait of patients suffering from Parkinson’s disease and normal gait, solely with the 2D walking videos recorded by a common camera or a smartphone. A pose estimation algorithm is employed to extract the skeleton of the subjects from the videos. Based on the analysis of motor disturbances resulting from Parkinson’s disease, specific gait features are defined and extracted, including step length, walking speed, arm swing magnitude, and velocity. Considering that the sample size of clinical data is limited in the early stage, classic classifiers are applied, including logistic regression (LR), support vector machine (SVM), and random forest (RF). The registered clinical study was conducted with 20 PD patients and 20 age-matched healthy controls. With the three classifiers, 87.5% (LR), 90.0% (SVM), and 92.5% (RF) of classification accuracy were achieved, respectively. Thus, video-based classification of abnormal gait promises a solution for remote screening and diagnosis of neurological diseases.

The Logit model and the Probit model are statistical models of Logistic regression and probability-based pattern recognition algorithms. The cases in which the Logit model and the Probit model fit better, respectively, are assessed to provide selection suggestions. Criteria, such as the AIC, Deviance, and Pseudo R-square, are selected to assess the goodness-of-fit of the two models. Using Monte Carlo simulation in different data scenarios found that some criteria have limited applicability in assessing the goodness of fit and two models are not sensitive to whether or not the independent variables are independent of each other. The difference between the two models is mainly in the sample size and the kurtosis of the data. The Probit model fits well to the smaller data set generated for all kurtosis cases, while the Logit model fits better when the sample size from the Leptokurtic distribution is larger.

Neural networks have shown remarkable classification performances in recent years, often outperforming humans in many tasks. Unfortunately, there are some tasks where conventional neural networks and their training methods have under-performed. One of these areas is learning from a small number of samples which have been partially addressed with the development of few-shot neural networks. Few-shot neural networks contrast traditional classification networks by learning classification tasks with datasets with only a few samples per class; however, these few-shot techniques are trained collectively with a large number of labeled samples. Hence, few-shot learning approaches only address the low sample per class learning problem whereas the task of learning strictly from a low sample size still goes mostly unresolved. In this contribution, we address the challenge of learning from high dimensional low sample data by revising the problem into a data ordering task. Specifically, we have designed OrderNet, a novel network design and training approach that can take a relatively small amount (less than 200 samples) of ordered high dimensional low sample data and organize many more unseen samples. To the best of our knowledge, OrderNet is the first neural network to address the high dimensional low sample data using techniques adopted from few-shot learning. We evaluate OrderNet against its ability to order images of analog clocks by time as well as images of profile pictures by age. Additionally, we demonstrate that OrderNet has superior performance over a conventional regression neural network in the low sample regime.

Deep learning-based plant disease management became a cost-effective way to improved agro-productivity. Advanced train sample generation and augmentation methods enlarge train sample size and improve feature distribution but generation and augmentation introduced sample feature discrepancy due to the generation learning process and augmentation artificial bias. We proposed a generation and geometric transformation invariant feature learning method using Siamese networks with maximum mean discrepancy loss to minimize the feature distribution discrepancies coming from the generated and augmented samples. Through variational GAN and geometric transformation, we created four dataset settings to train the proposed approach. The abundant evaluation results on the PlantVillage tomato dataset demonstrated the effectiveness of the proposed approach for the ResNet50 Siamese networks in learning generation and transformation invariant features for plant disease classification.

We study the problem of estimating properties of dynamic graphical models in the non-asymptotic regime. Instead of characterizing the behavior of estimation rules with asymptotic consistency, we study sharp bounds on the sample size required to estimate the properties. We show that for certain spatio-temporal Markov random fields governed by an underlying graph, one can estimate certain natural properties with logarithmic (w.r.t. the size of the underlying graph) sample complexity. Matching lower bounds are also established for such estimation problems. We highlight our results with a “bit river” abstraction, where a Bernoulli source at one node flows along the edges of an underlying graph, and the task is to obtain the flow trajectory. If we do not have any restriction on the model class under consideration, we also show that an exponential sample size is required for even very simple properties.

We consider the use of the spatio-temporal sample autocorrelation matrix in order to determine whether the signals received by a distributed antenna system are spatially correlated. The asymptotic behavior of linear spectral statistics built from this matrix is studied, assuming that (i) the number of antennas, (ii) the sample size and (iii) the number of tested time lags all converge to infinity. In this asymptotic regime, linear spectral statistics of the spatiotemporal sample correlation matrix are shown to be asymptotically equivalent to the corresponding functional average with respect to a Marchenko-Pastur distribution. This means that the eigenvalue distribution of the original sample autocorrelation matrix essentially behaves as a sample covariance matrix of spatio-temporal white noise with equivalent dimensions. This result turns out to be useful in order to address the problem of detecting the presence of wideband directional signals with a number of uncalibrated receivers distributed over a large area.

According to Diagnostic and Statistical Manual of Mental Disorders, Autism spectrum disorder (ASD) is a developmental disorder characterised by reduced social interaction and communication, and by restricted, repetitive, and stereotyped behaviour. An important characteristic of autism, referred in several diagnostic tests, is a deficit in eye gaze. The objective of this study is to review the literature concerning machine learning and eye-tracking in ASD studies conducted since 2015. Our search on PubMed identified 18 studies which used various eye-tracking instruments, applied machine learning in different ways, distributed several tasks and had a wide range of sample sizes, age groups and functional skills of participants. There were also studies that utilised other instruments, such as Electroencephalography (EEG) and movement measures. Taken together, the results of these studies show that the combination of machine learning, and eye-tracking technology can contribute to autism identification characteristics by detecting the visual atypicalities of ASD people. In conclusion, machine learning and eye-tracking ASD studies could be considered a promising tool in autism research and future studies could involve other technological approaches, such as Internet of Things (IoT), as well.

As a kind of microwave remote sensing imaging radar, Synthetic Aperture Radar (SAR) has been widely used in military and civil fields because of its advantages of all-weather and all-time imaging. Data is the "crude oil" to promote the development of deep learning. Aiming at the problem of how to use SAR data efficiently, this paper studies the data samples from different data source, different target sizes and different training sample sizes. Firstly, based on the SAR ship dataset, the expansion of specific samples is completed to increase the diversity of samples used in the experiment. Secondly, through the cross experiment of two data source, Sentinel-1 data has strong universality for different data source models. Further research on GF-3 data verifies that low resolution and small-scale target detection has great challenges, which provides some reference for researchers.

Direction-of-arrival (DOA) estimation problem is a challenging one in the presence of coherent sources, when the sample size is small, and the signal-to-noise ratio is low. We address this problem by developing a new method called enhanced standard ESPRIT (ES ESPRIT), and also its unitary extension called enhanced unitary ESPRIT (EU ESPRIT). The proposed methods use statistics of the subspace perturbation. First, they generate 2K DOA candidates for K sources, and then discreetly select K of them. Numerical results show the superiority of EU ESPRIT over other existing methods especially in improving threshold performance and separating closely located sources with a small sample size.

In this paper, the outlier production mechanism of the G-MUSIC Direction-of-Arrival estimation technique is investigated using tools from Random Matrix Theory. The G-MUSIC Direction-of-Arrival estimation technique is an improved version of the conventional MUSIC method that provides superior performance in low sample size scenarios. The stochastic behavior of the G-MUSIC cost function is analyzed in the asymptotic regime where both the number of snapshots and the number of antennas increase without bound at the same rate. The finite dimensional distribution of the G-MUSIC cost function is shown to be asymptotically jointly Gaussian. Furthermore, the probability of resolution of the G-MUSIC Direction-of-Arrival estimator is characterized by means of the derived asymptotic probability density function of the G-MUSIC cost function.

The objective of this study is to investigate and explore possible reasons that influence the intention to use social media for traveling in Gen Z. The conceptual framework is developed by proofed theories, Technology acceptance model (TAM model) and Theory of planned behavior (TPB). The result of sample size was 213 respondents whose ages range from 18 to 24 (Generation Z), the sampling method was a purposive sampling technique. The result indicates that perceived utility and perceived ease of use have statistically significant positive effect on consumers’ attitude. However, only perceived behavioral control, and subjective norms have a statistically significant positive effect on consumers’ intention to use social media for traveling in Gen Z. This concludes that the respondents are more likely to use social media for traveling, if they are more influenced by subjective norms such as their family agree for using social media for traveling, or friends recommend them to use it, or to follow their close-relationship person. Also, if respondents can manage the way to use social media on their demand, they are more likely to use social media for traveling.

Radio equipment for aircraft flight supporting consists of communication, navigation and surveillance devices. These devices are used to provide the safety and regularity of aircraft flights. During radio equipment operation the processes of intended use, maintenance, repair, continuation of life service are implemented. The decision-making about control and preventive actions for these processes is based on the results of statistical data processing. In such cases statistical methods with predetermined sample size are used. This paper concentrates on the problem of synthesis of sequential method that allows making decision faster for given level of data processing effectiveness. The researched sequential method is used for estimation of reliability parameters based on priory information about these parameters distribution and functional transformation of corresponding probability density function. During solving analysis problem, the probability density functions of estimate and observation duration are calculated.

Recognition methods of dangerous rhythm disturbances using the spectral description of short electrocardiosignals are the subject of this paper. Each object is described by 15 spectral features in the range from 0 to 15 Hz, all data are normalized to total signal strength. The sample size is 270 objects of each class. The method of Fisher linear discriminant (FLD) is used, as well as the method of convex hulls classification. The results of the binary classification of arrhythmias into non-dangerous and dangerous using two methods are presented. The performances of the classification efficiency (sensitivity, specificity, and overall accuracy) are calculated. Leave-one-out cross-validation (LOO CV) is used during the classification experiment to empirically evaluated the generalization ability of the algorithm. Also, mapping of two classes' objects onto the plane of vectors, obtained using the FLD method is presented. The convex hull classification based on linear programming has an overall accuracy of 91.5%. The results of this work can be used to solve practical problems in medicine.

Management Education Systems are developing yearly. This study is focusing whether achieving MBA degree brings any entrepreneurial qualities among students. The study is conducted in the colleges of Koehi. Five highly preferred colleges by the MBA students were selected for the study. The sample size of the study is 120 students with convenient random sampling. A well-structured interview cum questionnaire method is used for the study. The study reveals that the Management Education System creates awareness about the entrepreneurship. It also finds that certain qualities like innovativeness and creativeness, time management and many other traits are created by achieving the education. The tools selected for the study are Simple Average Method and Garrett Ranking system.

The aim of this work is to classify the dementia stage using microstructural white matter (WM) diffusion kurtosis indices. We develop various class imbalance machine learning methods in classifying the cognitive normals (CN), Alzheimer's disease (AD) and mild cognitive impairment (MCI) individuals. Diffusion-weighted images of 155 subjects were collected with 44 CN, 88 MCI and 23 AD individuals aged between 60 to 96 years. We first perform skull striping using FSL tool along with head motion and artifact correction. We calculate the mean diffusivity, fractional anisotropy from various brain regions of WM indices from diffusion kurtosis imaging (DKI) using tract based spatial statistics. Then we estimate the kurtosis fractional anisotropy of various WM regions. After estimation of all the regional KFA indices, we trained these indices using various class imbalance machine learning classification algorithms, as the sample size is different in all the three groups. We found that balanced random forest classifier was the best classifier with an accuracy of 74% when compared to other methods for 5-fold cross-validation. We conclude that class imbalance machine learning methods are potential in classifying cognitive normals, AD and cognitive impairment individuals.

Amidst rising rates of adolescent obesity and overuse of digital technologies amongst secondary school students, educators and policymakers are especially keen on boosting sports participation amongst Chinese youth. Furthermore, the existence of a robust relationship between sports participation and academic performance would be an additional reason to expand sports-related activities in school. The Ministry of Education of China is also calling for attention for Physical Education in recent years. In statistics, ordinary least squares (OLS) is a type of linear least squares method for estimating the unknown parameters in a linear regression model. It is widely used in social sciences as it is easy to understand and calculate. In fact, numerous studies have found that sports participation does have a positive effect on students’ academic performance based on regression model. By using the OLS and interaction effect on the data of China Education Panel Survey (CEPS) between the 2013 - 2014 academic years, this paper use data study correlations between sports participation and academic performance. Based on a sample size of 4033 secondary school students from China, this paper shows that there is a negative correlation between students’ sports intensity and exam scores. Also, this study found that sex plays an interaction effect on this negative correlation. Findings of this paper indicate that the lack of a comprehensive and robust Physical Education program is a weakness of the current Chinese education.

The steel industry is one of the pillar industries in China. The physical and mental health of steel workers is related to the development of China's steel industry. Steel workers have long been working in shifts, high temperatures, noise, highly stressed, and first-line environments. These occupational related factors have an impact on the health of steel workers. At present, the existing hypertension risk scoring models do not include occupational related factors, so they are not applicable to the risk score of hypertension in steel workers. It is necessary to establish a risk scoring model for hypertension in steel workers. In this study, the learning vector quantization (LVQ) neural network algorithm and the FisherSVM coupling algorithm are applied to estimate the hypertension risk of steel workers, and the microscopic laws of the "tailing" phenomenon of the two algorithms are analyzed by means of graphics analysis, which can describe the influence trend of sample size change in different intervals on the classification effect. The results show that the classification accuracy of the algorithm depends on the size of the sample space. When the sample size n ≤ 30 * (k + 1), the Fisher-SVM coupling intelligent algorithm is more applicable. Because its average accuracy rate is 90.00%, the average accuracy of the LVQ algorithm is only 63.34%. When the sample size is n > 30 * (k + 1), the LVQ algorithm is more applicable. Because its average accuracy rate is 93.33%, the average accuracy of the Fisher-SVM coupling intelligent algorithm is only 76.67%. The sample size of this paper is 4422, and the prediction of LVQ neural network model is more accurate. Therefore, based on the relative importance of each risk factor obtained by this model and to establish a steel worker hypertension risk rating scale, the score greater than 18 is considered as the high risk, 12-18 is considered as the medium risk, and less than 12 is considered as the low risk. Through the example's verification, the accuracy rate of the scale is 90.50% and the effect is very good. It shows that the established scoring system can effectively assess the risk of hypertension in steel workers and provide an effective basis for primary prevention of hypertension in steel workers.

In this paper, a process yield-index for autocorrelation between nonlinear profiles is proposed. Applying a one-dimensional Taylor series expansion, the mean and variance of the estimator of the yield index are derived. To evaluate the performance of the proposed yield index a simulation study is conducted. The new index is employed to design three acceptance sampling plans for quality characteristics described by auto-correlated nonlinear profiles. The first sampling plan is based on resubmission. The remaining two follow the repetitive group and multiple dependent repetitive sampling schemes respectively. For all sampling plans, the operating characteristic function is developed. A non-linear optimization approach with search algorithm is employed to determine the number of profiles required for inspection and to decide critical values for acceptance, rejection, and resubmission criteria. The performance of the new methods is investigated and compared with traditional single sampling plan. The comparison confirms all of the three proposed methods possess higher efficiency than a single sampling plan in terms of sample size. For practical applications tables are provided. Two numerical examples from the automobile engine testing and particleboard manufacturing process are used to demonstrate the applicability of the proposed methods.

High-throughput genomic technologies are leading to a paradigm shift in research of computational biology. Computational analysis with high-dimensional data and its interpretation are essential for the understanding of complex biological systems. Most biological data (e.g., gene expression and DNA sequence data) are high-dimensional, but consist of much fewer samples than predictors. Such high-dimension, low sample size (HDLSS) data often cause computational challenges in biological data analysis. A number of least absolute shrinkage and selection operator (LASSO) methods have been widely used for identifying biomarkers or prognostic factors in the field of bioinformatics. The LASSO solution has been improved through the development of the LASSO derivatives, including elastic-net, adaptive LASSO, relaxed LASSO, VISA, random LASSO, and recursive LASSO. However, there are several known limitations of the existing LASSO solutions: multicollinearity (particularly with different signs), subset size limitation, and the lack of the statistical test of significance. We propose a high-dimensional LASSO (Hi-LASSO) that theoretically improves a LASSO model providing better performance of both prediction and feature selection on extremely high-dimensional data. The Hi-LASSO alleviates bias introduced from bootstrapping, refines importance scores, improves the performance taking advantage of global oracle property, provides a statistical strategy to determine the number of bootstrapping, and allows tests of significance for feature selection with appropriate distribution. The performance of Hi-LASSO was assessed by comparing the existing state-of-the-art LASSO methods in extensive simulation experiments with multiple data settings. The Hi-LASSO was also applied for survival analysis with GBM gene expression data.

The Malaysian women population frequently face the problem of finding the best fitting shoes. This problem is created by the absence of a Malaysian women shoe sizing system. Standard statistical methods involving the Multivariate Normal distribution are used in a novel process of addressing issues related to the creation of a shoe sizing system, in particular, the problem of defining categories of shoe sizes. This study focused on the use of five-foot measurement namely, foot length (FL), foot breadth (FB), foot's ball girth (BG), instep length (IL), and fibulare instep length (FIL). Univariate hypothesis testing was performed taking advantage of the existence of normal probability distribution. For brevity, details for FL, FB, and BG are shown in this paper, followed by a comparison of performance results between (FL, FB, BG) and (FL, FB, BG, IL, FIL). Our results were compared to a similar study showing almost the same aggregate loss and coverage percentage. The result shows that a modest sample size of 160 was sufficient to define categories of shoe sizes to help develop a prototype shoe sizing system using the proposed novel approach. The proposed prototype shoe sizing system provides information for the planning, design, and manufacturing of Malaysian women's footwear with implications for better fitness and comfort.

Locating Lingwu Long Jujubes is a key step for the automatic picking of the jujubes which can lower labour costs. In this paper, a method for detecting Lingwu Long Jujubes in a natural environment with a three-dimensional (3D) point cloud is proposed. First, the jujubes are preliminarily extracted in two-dimensional (2D) image. Then, the points are fitted to ellipsoid by least square (LS) in the sample consensus framework. The model scores are different when sample size is changed. The variable sample consensus (VARSAC) is proposed to get higher score than random sample consensus (RANSAC). The normal vector and the distance need to be calculated in score calculation of the RANSAC. However, this score calculation method is complex and time-consuming. Thus, a new method called two-ellipsoid-bounding-counting (TEBC) is proposed. The TEBC produces two auxiliary ellipsoids that are obtained by scaling semiaxis of the model. The points, which is bounded by two auxiliary ellipsoids, are regarded as inliers. The functional value of every candidate point is calculated to select the inliers. The valid ellipsoids are determined by the prior information and the invariants. Finally, the centre, size and the attitude angle of the jujubes are solved using eigenvalues and eigenvectors. Experiments are carried out on synthetic and real datasets. The experimental results show that the proposed method can faster and more accurately detect jujube. The speed of the VARSAC+TEBC is approximately 4 times faster than that of the RANSAC in the real dataset.

Cluster analysis is a multivariate data mining technique that is widely used in several areas. It aims to group automatically the n elements of a database into k clusters, using only the information of the variables of each case. However, the accuracy of the final clusters depends on the clustering method used. In this paper, we present an evaluation of the performance of main methods for cluster analysis as Ward, K-means, and Self-Organizing Maps. Differently from many studies published in the area, we generated the datasets using the Design of Experiment (DOE) technique, in order to achieve reliable conclusions about the methods through the generalization of the different possible data structures. We considered the number of variables and clusters, dataset size, sample size, cluster overlapping, and the presence of outliers, as the DOE factors. The datasets were analyzed by each clustering method and the clustering partitions were compared by the Attribute Agreement Analysis, providing invaluable information about the effects of the considered factors individually and about their interactions. The results showed that, the number of clusters, overlapping, and the interaction between sample size and number of variable significantly affect all the studied methods. Moreover, it is possible to state that the methods have similar performances, with a significance level of 5%, and it is not possible to affirm that one outperforms the others.

Both wrapper and hybrid methods in feature selection need the intervention of learning algorithm to train parameters. The preset parameters and dataset are used to construct several sub-optimal models, from which the final model is selected. The question is how to evaluate the performance of these sub-optimal models? What are the effects of different evaluation methods of sub-optimal model on the result of feature selection? Aiming at the evaluation problem of predictive models in feature selection, we chose a hybrid feature selection algorithm, FDHSFFS, and conducted comparative experiments on four UCI datasets with large differences in feature dimension and sample size by using five different cross-validation (CV) methods. The experimental results show that in the process of feature selection, twofold CV and leave-one-out-CV are more suitable for the model evaluation of low-dimensional and small sample datasets, tenfold nested CV and tenfold CV are more suitable for the model evaluation of high-dimensional datasets; tenfold nested CV is close to the unbiased estimation, and different optimal models may choose the same approximate optimal feature subset.

Experiments used for model validation constitute a new type of experiments, whose primary goal is to determine the ability of a high-fidelity mathematical model in simulating a well-characterized physical process. Considering the importance of the credibility of the validation experiment, in this paper, a methodology for the optimal design of validation experiments based on the area metric factor and the fuzzy expert system is presented to obtain the test scheme with low cost and satisfactory credibility. First, the concept of area metric factor is proposed, which is a dimensionless validation metric. Then, a criterion for qualitative assessment of the predictive model accuracy is developed based on the fuzzy expert system, in which the inconsistency among the different expert groups is considered. Subsequently, an optimization model of the validation experiment design is constructed, in which the sample size of the experiments is designed variable, while the credibility of the validation experiment is a constraint. Meanwhile, a novel method is developed to solve the optimization model, in which the Latin hypercube sampling is used to obtain the location of each validation experiment. Finally, two simulation examples are adopted to validate the proposed methods. The simulation results show that the randomness of the experimental scheme significantly affects the credibility of the evaluation results in the case of small sample sizes. The proposed method for the validation experiment design can reduce the impact of the selection of the testing programs on the credibility of model validation.

Determining the required probe vehicle size for real-time travel time estimation is an important issue in probe-data-based applications, such as traffic monitoring on a roadway network. This paper provides an innovative approach for the determination of proper sample sizes with the aid of a simulation model. The model, which was built for El Camino Real, a major arterial in the California San Francisco Bay Area, helps to provide the necessary detailed data to verify the proposed methodology. Our proposed approach is based on the findings that the distribution of travel time estimate error can be constructed with the generalized Pareto distribution function. Considering that travel times vary within a signal cycle and among different movements (through, left-turning and right-turning) at intersections, our study uses data in 15-s intervals to examine the travel time variations associated with different intersection movements. Our findings show that when the penetration rate of probe vehicles is 40%, the probability that the relative error will be less than 0.10 within the time period is 90%, whereas a penetration rate greater than 60% is needed to attain a 0.95 confidence level at the same error rate. For left-turning vehicles, a 10% penetration of probe vehicles is sufficient for estimating their travel times with relative errors less than 0.10 at a 90% confidence level, whereas through vehicles and right-turning vehicles require substantially higher penetration rates of approximately 50% and 30%, respectively, to attain the same confidence level.

Current group dynamics and cognitive-developmental theories highlight the significance of active participation in the learning environment. Whereas on-going learning practices suggest that for creating a better learning environment, the most important step is to construct learning as an active and joint process carried under safe circumstances. Specifically, psychological safety plays a crucial role in activating knowledge and learning within organizations at both individual and group levels. This research has measured the relationship between psychological safety, facilitation, and teamwork results such as the team’s performance across an educational environment. A closed-ended questionnaire has been used to collect the data. The sample size is ninety students who participated in an organized workshop. The statistical tool SPSS version 12 (Mann-Whitney U-test) is used to examine the data and interpret the results. Based on the outcomes, psychological safety and facilitation approach in combination are proposed to enrich the teamwork efficiency in an academic sector.

Obese people have high chances of cardiovascular disease (CVD), which is supposed to be due to the alteration in autonomic nervous system (ANS) activity. The changes in ANS activity can be identified using heart rate variability (HRV). HRV is a non-invasive tool to measure the ANS activity using linear and non-linear HRV features. The paper presents an aim to understand the effect of obesity on ANS using HRV parameters. Initially, sixteen control and sixteen obese subjects of both the gender between ages 20 to 50 were involved in the study after that synthetic minority oversampling technique (SMOTE) was used to increase the sample size of control and obese subjects from sixteen to forty-eight. The statistically significant difference between two groups was observed using the Independent t test. The statistical results of the study indicate the sympathovagal imbalance due to reduced parasympathetic activity. The statistical results were validated by incorporating the machine learning technique into the study. Machine Learning (ML) algorithm helps to identify the most important predictor that can clearly differentiate control and obese subjects. The statistical and ML algorithm result shows changes in the sympathovagal balance due to decreased parasympathetic activity.

The Fisher information matrix provides a way to measure the amount of information given observed data based on parameters of interest. There are many relative applications in statistical modeling, system identification, and parameter estimation. This article presents an enhanced resampling-based method with independent perturbation to estimate the Fisher information matrix. We show its accuracy via variance reduction that is reduced by a factor of $n$ , where $n$ is the sample size, with a numerical example under the signal-plus-noise setting also illustrating same.

When doing multivariate data analysis, one common obstacle is the presence of incomplete observations, i.e., observations for which one or more key fields are missing data. Missing data is often countered by deleting entire observations that contain missing data. The negative effects of deleting entire observations are multiple: deleting observations reduces sample size and can also result in biased inferences even if data is missing at random. In addition, knowledge contained within incomplete observations is knowledge lost when they are deleted- and the effort spent collecting that knowledge is effort wasted. Data imputation methods, or methods of statistically “filling-in” missing data, can help combat small sample sizes by using the existing information in partially complete observations with the end goal of producing less biased and higher confidence inferences. When a sample from a multivariate normal population is only partially complete, and the missing data meets appropriate assumptions (missing at random), robust data imputation of the missing data can be implemented with monotone data augmentation (MDA) using the multivariate t distribution. Missing data imputation is applied to data from the NASA Instrument Cost Model (NICM) using the MDA algorithm under the assumption of having a multivariate t distribution with fixed degrees of freedom. A sensitivity analysis to the degrees of freedom parameter is presented to demonstrate robustness of the multivariate t distribution when dealing with small samples as compared to the multivariate normal distribution.

Motor imagery brain computer interface designs are considered difficult due to limitations in subject-specific data collection and calibration, as well as demanding system adaptation requirements. Recently, subject-independent (SI) designs received attention because of their possible applicability to multiple users without prior calibration and rigorous system adaptation. SI designs are challenging and have shown low accuracy in the literature. Two major factors in system performance are the classification algorithm and the quality of available data. This paper presents a comparative study of classification performance for both SS and SI paradigms. Our results show that classification algorithms for SS models display large variance in performance. Therefore, distinct classification algorithms per subject may be required. SI models display lower variance in performance but should only be used if a relatively large sample size is available. For SI models, LDA and CART had the highest accuracy for small and moderate sample size, respectively, whereas we hypothesize that SVM would be superior to the other classifiers if large training sample-size was available. Additionally, one should choose the design approach considering the users. While the SS design sound more promising for a specific subject, an SI approach can be more convenient for mentally or physically challenged users.

A notable problem in Brain-Computer Interface (BCI) is the burden of collecting an adequate amount of training data to be used in estimating a robust classifier model. This study explores the prospect of using Generative Adversarial Networks (GANs), a novel family of generative models, to perform data augmentation for generating artificial training data that are used in the classification of P300 event-related potentials in electroencephalogram (EEG) signals. In this regard, we consider two popular GANs, namely, Deep Convolutional Generative Adversarial Networks (DCGAN) and Wasserstein GAN with Gradient Penalty (WGAN-GP), and explore their efficacy in generating artificial EEG data in both subject-specific and subject-independent contexts. The results show that data augmentation using both DCGAN and WGAN-GP could lead to improved classification performance. However, the operating conditions that an improvement is observed differ between the subject-specific and subject-independent classification schemes. In particular, we observe that a transition from a relatively small to large training sample size per subject would generally lead to a better classification performance in training subject-specific classifiers; however, when a limited number of subjects is available, this transition could potentially result in an opposite effect in case of subject-independent classification.

Sex identification plays an important role to identify individuals in a forensic investigation when unknown body parts are found. This study aims to use machine learning (ML) to determine sex using different hand dimensions in the Bangladeshi population. Nine hand measurements such as hand length, handbreadth, maximum handbreadth, palm length, thumb finger length, index finger length, middle finger length, ring finger length, and little finger length are examined to identify the sex. 292 Bangladeshi adults (145 males and 147 females) aged between 18 and 60 years old were considered as sample size, while two ML approaches such as the linear discriminant analysis and logistic regression methods are used to estimate individual sex. The ML approaches are validated using a 10-fold cross-validation scheme. Varying accuracy was found using different algorithms and different combinations of features – one feature at a time and all the features altogether. While all the features were considered altogether, for left-hand measurements, the logistic regression algorithm demonstrated maximum accuracy (91.10%), whereas for right-hand measurements altogether linear discriminant analysis algorithm showed a maximum accuracy of 91.40%. As a single feature, palm length demonstrated the maximum accuracy in both hands for all the algorithms.

This paper establishes the Firefly Algorithm (FA) to be a special variant of the conventional Accelerated Particle Swarm Optimization (APSO) technique by suggesting modifications in the movement criteria of the fireflies in case of the simple FA. In order to find the significance of this implication, a well known optimization problem known as the Economic Dispatch problem is computed for multiple test cases. Economic dispatch problem aims to minimize the total fuel cost of a multi-generator power system. In addition, the valve point effect loading is considered for the thermal cost equation in order to make the objective function more non-linear and non-convex in nature. The modified and enhanced FA not only improves its performance as compared to the conventional FA, but also presents enhanced FA as a special case of the APSO algorithm by giving the similar performance parameters as that of APSO. Moreover, a comprehensive statistical analysis based on the results of the independent t-test is presented in order to statistically compare the performance of APSO, FA and enhanced FA. The independent t-test results statistically prove enhanced FA to be a special variant of APSO technique by comparing the mean and the variance of the two algorithms for a particular sample size.

Safety-critical systems, such as medical products, industrial safety functions, or autonomous driving systems, rely not only on the knowledge of the actual system parameters, but it is imperative to also take statistic properties into account. Besides measurement uncertainties, sample variation can play an extraordinary role in the evaluation of the overall variation of a certain parameter. S-parameters are used to describe the linear behavior of high-frequency devices, such as cables. This paper focuses on the quantification of sample variation to satisfy predefined safety margins. First, statistic relations are deduced and presented. Afterwards, these results are verified by means of Monte Carlo simulations. It can be shown, that even for moderate sample sizes of about 50 observations, the Hotelling’s T2 distribution needs to be used to account for the uncertainty of the sample covariance matrix estimation. These general findings are adapted to S-parameter measurements and an application based on 9 cable measurements is presented.

Based on test data of physical form of women’s volleyball team members who participated in "National Elite Young Female Volleyball Training Program" from 2011 to 2018, applying information technology, this study has grouped data according to each year and each age group and therefore obtained statistical data as well as established databases on the average, mean value, minimum value, maximum value as well as sample size of the athletes’ height(cm), weight(kg), BMI, hand-foot spacing(cm), finger spacing(cm). Applying the Excel software, this study has arranged and compared the average data of these five indicators, systemically analyzed the advantages and disadvantages of physical form of each age group from each year, thus revealing the characteristics of physical form of our country’s high-level reserve talents of women’s volleyball team, in an attempt to make sensible selections of teenage athletes, improve the rate of talent as well as to provide a theory basis for a scientifically appropriate training regime.

Facial expression recognition (FER) is widely studied in the area of man-machine interaction. But in the current research on expression recognition technique, the degradation of recognition performance is caused by the problem of small sample and unbalance of sample size in expression database. In this paper, we study the generation of facial expression image in generative adversarial model, focusing on the solution to meet the Lipschitz constraint. We propose to generate the facial expression images based on spectral norm normalized networks (SNN-EG). The method satisfying K-Lipschitz regular constraints, can expand the number of expression samples, enrich sample diversity, and achieve the solution for the problems of small samples and unbalanced samples. The KDEF expression dataset is validated and evaluated. Experiments show that the proposed method has a better effect on synthesis and recognition of facial expressions.

Most causal learning algorithms, especially the ones using conditional independence tests, suffer in the low sample regime. In this paper, we examine the possibility of causal discovery using only the pairwise dependency structure of a domain and investigate if this approach can lead to more robust and reliable solutions to causal discovery when the sample size is limited. We show that marginal independence relations can be used to construct a novel causal discovery tool capable of distinguishing between direct and indirect causal effects and detecting a lower bound on direct causal effects. In our simulations we found that causal discovery using marginal independency information can be significantly more accurate than causal knowledge obtained through conditional independence tests when the sample size is limited: our experiments suggests avoiding conditional independence tests can reduce error rate by up to 30 percent. Finally, using this marginal independency based causal discovery tool, we propose four candidate genes that possibly contain the causal mutation in Major Depressive Disorder utilizing a database of only 59 samples.