Size Of Sample Set Research Articles

Evaluation Methodologies in Software Protection Research

Man-at-the-end (MATE) attackers have full control over the system on which the attacked software runs, and try to break the confidentiality or integrity of assets embedded in the software. Both companies and malware authors want to prevent such attacks. This has driven an arms race between attackers and defenders, resulting in a plethora of different protection and analysis methods. However, it remains difficult to measure the strength of protections because MATE attackers can reach their goals in many different ways and a universally accepted evaluation methodology does not exist. This survey systematically reviews the evaluation methodologies of papers on obfuscation, a major class of protections against MATE attacks. For 571 papers, we collected 113 aspects of their evaluation methodologies, ranging from sample set types and sizes, over sample treatment, to performed measurements. We provide detailed insights into how the academic state of the art evaluates both the protections and analyses thereon. In summary, there is a clear need for better evaluation methodologies. We identify nine challenges for software protection evaluations, which represent threats to the validity, reproducibility, and interpretation of research results in the context of MATE attacks and formulate a number of concrete recommendations for improving the evaluations reported in future research papers.

Split-sample reliability estimation in health care quality measurement: Once is not enough.

To examine the sensitivity of split-sample reliability estimates to the random split of the data and propose alternative methods for improving the stability of the split-sample method. Data were simulated to reflect a variety of real-world quality measure distributions and scenarios. There is no date range to report as the data are simulated. Simulation studies of split-sample reliability estimation were conducted under varying practical scenarios. All data were simulated using functions in R. Single split-sample reliability estimates can be very dependent on the random split of the data, especially in low sample size and low variability settings. Averaging split-sample estimates over many splits of the data can yield a more stable reliability estimate. Measure developers and evaluators using the split-sample reliability method should average a series of reliability estimates calculated from many resamples of the data without replacement to obtain a more stable reliability estimate.

Overcoming data scarcity in radiomics/radiogenomics using synthetic radiomic features

PurposeTo evaluate the potential of synthetic radiomic data generation in addressing data scarcity in radiomics/radiogenomics models. MethodsThis study was conducted on a retrospectively collected cohort of 386 colorectal cancer patients (n = 2570 lesions) for whom matched contrast-enhanced CT images and gene TP53 mutational status were available. The full cohort data was divided into a training cohort (n = 2055 lesions) and an independent and fixed test set (n = 515 lesions). Differently sized training sets were subsampled from the training cohort to measure the impact of sample size on model performance and assess the added value of synthetic radiomic augmentation at different sizes. Five different tabular synthetic data generation models were used to generate synthetic radiomic data based on “real-world” radiomics data extracted from this cohort. The quality and reproducibility of the generated synthetic radiomic data were assessed. Synthetic radiomics were then combined with “real-world” radiomic training data to evaluate their impact on the predictive model's performance. ResultsA prediction model was generated using only “real-world” radiomic data, revealing the impact of data scarcity in this particular data set through a lack of predictive performance at low training sample numbers (n = 200, 400, 1000 lesions with average AUC = 0.52, 0.53, and 0.56 respectively, compared to 0.64 when using 2055 training lesions). Synthetic tabular data generation models created reproducible synthetic radiomic data with properties highly similar to “real-world” data (for n = 1000 lesions, average Chi-square = 0.932, average basic statistical correlation = 0.844). The integration of synthetic radiomic data consistently enhanced the performance of predictive models trained with small sample size sets (AUC enhanced by 9.6%, 11.3%, and 16.7% for models trained on n_samples = 200, 400, and 1000 lesions, respectively). In contrast, synthetic data generated from randomised/noisy radiomic data failed to enhance predictive performance underlining the requirement of true signal data to do so. ConclusionSynthetic radiomic data, when combined with real radiomics, could enhance the performance of predictive models. Tabular synthetic data generation might help to overcome limitations in medical AI stemming from data scarcity.

Uncertainty Quantification in Heterogeneous Treatment Effect Estimation with Gaussian-Process-Based Partially Linear Model

Estimating heterogeneous treatment effects across individuals has attracted growing attention as a statistical tool for performing critical decision-making. We propose a Bayesian inference framework that quantifies the uncertainty in treatment effect estimation to support decision-making in a relatively small sample size setting. Our proposed model places Gaussian process priors on the nonparametric components of a semiparametric model called a partially linear model. This model formulation has three advantages. First, we can analytically compute the posterior distribution of a treatment effect without relying on the computationally demanding posterior approximation. Second, we can guarantee that the posterior distribution concentrates around the true one as the sample size goes to infinity. Third, we can incorporate prior knowledge about a treatment effect into the prior distribution, improving the estimation efficiency. Our experimental results show that even in the small sample size setting, our method can accurately estimate the heterogeneous treatment effects and effectively quantify its estimation uncertainty.

Abstract 6155: Germline variants in cancer susceptibility genes and subsequent neoplasm risks after childhood cancer: A pooled analysis of two large-scale cohorts

Abstract Background: Increasing evidence suggests germline genetic susceptibility influences the risk of developing subsequent neoplasms (SNs)—a major cause of morbidity and mortality—among childhood cancer survivors, but studies have been hampered by sample size and limited gene sets. We pooled two large-scale cohorts of childhood cancer survivors to characterize overall and specific SN risks among individuals carrying germline variants in 88 known cancer susceptibility genes with an autosomal dominant inheritance pattern. Methods: Using whole-genome or whole-exome sequencing data from the Childhood Cancer Survivor Study and St. Jude Lifetime Cohort (median [interquartile range] follow-up, 29.7 [13.8] years), we identified rare, potentially protein-damaging germline variants with SnpEff/ClinVar. Conditional logistic regression evaluated overall SN and SN-specific risk, with up to 100 SN-free controls matched to cases by age, sex, primary childhood cancer type, SN type, radiation dose, chemotherapy, study, and follow-up time. Results: Among 11840 individuals (50.6% female) who survived childhood cancer, 2162 (18.3%) developed at least one SN. The risk of developing any SN was elevated among 288 survivors carrying rare, potentially protein-damaging germline variants in an autosomal dominant cancer susceptibility gene (cases=13.3%; controls=9.6%; odds ratio [OR], 1.48; 95% confidence interval [CI], 1.30-1.70; P=5.12x10-5). A similar proportion of carriers was observed among those who developed common SNs such as basal cell carcinoma (N=107/882, 12.1%), meningioma (N=52/373, 13.9%), breast cancer (N=49/353, 13.9%), and thyroid cancer (N=32/243, 13.2%). Higher frequencies of carriers were observed in less common SNs, specifically sarcoma (N=24/151, 15.9%), colorectal cancer (N=15/72, 20.8%), and glioma (N=16/64, 25.0%). The observed associations were generally consistent regardless of treatment type for first primary childhood cancer. Gene-specific analyses revealed that the risk of developing any SN was most pronounced among individuals carrying variants in TP53 (cases=0.6%; controls=0.04%; OR, 9.93; 95% CI, 4.62-21.34; P=0.001) and FANCM (cases=0.9%; controls=0.5%; OR, 2.45; 95% CI, 1.48-4.03; P=0.018). Conclusion: In large-scale cohorts of childhood cancer survivors with long-term follow-up, individuals carrying potentially protein-damaging germline variants in known autosomal dominant cancer susceptibility genes have substantially increased SN risk, emphasizing the importance of genetic testing and consideration of genetic risk in follow-up guidelines. However, these individuals account for a small fraction of all SN cases, warranting future research on underlying risk factors in this population. Citation Format: Shahriar A. Zamani, Danielle M. Karyadi, Stephen W. Hartley, Joshua N. Sampson, Todd Gibson, Peter Kraft, Stephen J. Chanock, Lindsay M. Morton. Germline variants in cancer susceptibility genes and subsequent neoplasm risks after childhood cancer: A pooled analysis of two large-scale cohorts [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6155.

Analysis of the technical condition of gas turbine plants. Part 3. Application of machine learning methods

A method for estimating the technical condition coefficient of a gas turbine unit for transporting of natural gas using machine learning methods is considered. Archival gas-dynamic parameters recorded by the installation’s automatic control system are used as initial data. The initial data set was created by determining power from the change in enthalpy of natural gas before and after the blower. The software package is implemented in Python; the Scikit-learn library is used for machine learning models. The average absolute error in percentage is chosen as a criterion for the quality of the prediction. The prediction quality of machine learning models is assessed for different sets of parameters-features and sample sizes. Recommendations on the use of models are given. Keywords: gas turbine unit, technical condition, measuring system, data processing, machine learning. v.l.blinov@urfu.ru; deryabingleb.were@gmail.com

A rank-based approach to evaluate a surrogate marker in a small sample setting.

In clinical studies of chronic diseases, the effectiveness of an intervention is often assessed using "high cost" outcomes that require long-term patient follow-up and/or are invasive to obtain. While much progress has been made in the development of statistical methods to identify surrogate markers, that is, measurements that could replace such costly outcomes, they are generally not applicable to studies with a small sample size. These methods either rely on nonparametric smoothing which requires a relatively large sample size or rely on strict model assumptions that are unlikely to hold in practice and empirically difficult to verify with a small sample size. In this paper, we develop a novel rank-based nonparametric approach to evaluate a surrogate marker in a small sample size setting. The method developed in this paper is motivated by a small study of children with nonalcoholic fatty liver disease (NAFLD), a diagnosis for a range of liver conditions in individuals without significant history of alcohol intake. Specifically, we examine whether change in alanine aminotransferase (ALT; measured in blood) is a surrogate marker for change in NAFLD activity score (obtained by biopsy) in a trial, which compared Vitamin E ($n=50$) versus placebo ($n=46$) among children with NAFLD.

Risk Factors Associated with Pyogenic Spinal Infections among Intravenous Drug Users and Nonusers.

To identify the differences in patient factors, presentation, treatment course, and outcomes of intravenous drug users (IVDA) vs. nonusers (non-IVDA) presenting with pyogenic spinal infections. Study Design/Setting. Retrospective case series. We identified all cases involving spinal infections at our institution between May 2017 and January 2023. Postsurgical infections were excluded, and patients were separated into IVDA and non-IVDA groups. The patient charts were reviewed and analyzed for statistical or clinically significant differences using RStudio (2019 version 3.6.2). Our institutional review board approved this study, IRB# 2020-277, iRISID-2023-1384. Fifty patients (29 males and 21 females) with primary pyogenic spinal infections were included in our study. There were fourteen patients (28.0%) in the IVDA group. The mean age in the IVDA group was 50.6 vs. 61.9 years (p < 0.05) in the non-IVDA group. The average length of stay (LOS) in the IVDA group was 15.8 vs. 14.0 days (p = 0.54) in the non-IVDA group, with no significant difference in readmissions or disposition. Twenty-three non-IVDA patients were diagnosed with diabetes, while eight IVDA patients had a psychiatric diagnosis (other than substance abuse). There were no significant differences in microbial isolate or the duration of antibiotics. In our study population, there is a high incidence of intravenous drug abuse (IVDA), psychiatric disease, diabetes, and chronic kidney disease. Analysis shows a trend of two distinct patient populations. Patients without a history of IVDA were significantly older than those with IVDA and significantly more likely to have medical comorbidities including hyperlipidemia, diabetes, chronic kidney disease, and malignancy than those with IVDA history. Patients with IVDA were younger with significantly higher rates of smoking and psychiatric disorders. IVDA patients struggled to receive continued psychiatric/addiction treatment after discharge, an area for significant improvement. Due to a small sample size and single urban institution setting, this study may be underpowered to demonstrate differences in healthcare resource consumption.

Smart Multistage Privacy-Preserving Framework for Intrusion Detection in Multi-Domain SDN

SDN architectures are frequently used by organizations for the management of their networks and the detection of anomalous traffic in a single domain. However, in the real world, anomalous traffic might result in attacks like distributed denial of service (DDoS) that affect numerous domains. During intrusion detection, each SDN domain has to send real traffic data of a large volume to the multi-domain controller, exposing its sensitive information. This paper proposes a smart multistage framework for detecting attacks and ensuring privacy at no additional cost. This work utilized the recent unbalanced InSDN dataset for experimentation. It also uses an oversampling technique that reduces the imbalance rate for each attack type and selects the smallest possible training size and feature set size for an increase in detection accuracy and a reduction in computational complexity. Then, a multi-class classifier method for intrusion detection that does not require regularization or hyperparameter tuning, called ensemble-learning-based shallow decision tree (ELSDT) is proposed. Furthermore, the performance of the proposed classifier on the InSDN dataset is assessed on an SDN testbed. Experimental results show the ability of the proposed smart multistage privacy-preserving framework to make a significant reduction in the training sample size and feature set size to 87% and 76%, respectively. It also shows its outperformance in recent literature works by 5.67% improved accuracy.

Addressing sequential and concurrent treatment regimens in a small n sequential, multiple assignment, randomized trial (snSMART) in the MISTIC study

ABSTRACT Multisystem Inflammatory Syndrome in children (MIS-C) is a rare and novel pediatric complication linked to COVID-19 exposure, which was first identified in April 2020. A small n, Sequential, Multiple Assignment, Randomized Trial (snSMART) was applied to the Multisystem Inflammatory Syndrome Therapies in Children Comparative Effectiveness Study (MISTIC) to efficiently evaluate multiple competing treatments. In the MISTIC snSMART study, participants are randomized to one of three interventions (steroids, infliximab or anakinra), and potentially re-randomized to the remaining two treatments depending on their response to the first randomized treatment. However, given the novelty and urgency of the MIS-C disease, in addition to patient welfare concerns, treatments were not always administered sequentially, but allowed to be administered concurrently if deemed medically necessary. We propose a pragmatic modification to the original snSMART design to address the analysis of concurrent versus sequential treatments in the MISTIC study. A modified Bayesian joint stage model is developed that can distinguish a concurrent treatment effect from a sequential treatment effect. A simulation study is conducted to demonstrate the improved accuracy and efficiency of the primary aim to estimate the first stage treatments’ response rates and the secondary aim to estimate the combined first and second stage treatments’ responses in the proposed model compared to the standard snSMART Bayesian joint stage model. We observed that the modified model has improved efficiency in terms of bias and rMSE under large sample size settings.

Mixture of experts distributional regression: implementation using robust estimation with adaptive first-order methods

In this work, we propose an efficient implementation of mixtures of experts distributional regression models which exploits robust estimation by using stochastic first-order optimization techniques with adaptive learning rate schedulers. We take advantage of the flexibility and scalability of neural network software and implement the proposed framework in mixdistreg, an R software package that allows for the definition of mixtures of many different families, estimation in high-dimensional and large sample size settings and robust optimization based on TensorFlow. Numerical experiments with simulated and real-world data applications show that optimization is as reliable as estimation via classical approaches in many different settings and that results may be obtained for complicated scenarios where classical approaches consistently fail.

HLEL: A wetland classification algorithm with self-learning capability, taking the Sanjiang Nature Reserve I as an example

Wetland ecosystems have been severely damaged by atmospheric pollution and other environmental problems for decades. Reasonable protection, restoration, and utilization of wetlands have become priorities for environmental protection. Obtaining high-precision wetland map products is essential for ecological research as well as formulation and implementation of conservation policies. This study proposes a hierarchical labeling ensemble learning (HLEL) algorithm with self-learning capabilities to achieve high-precision wetland land cover and land use (LULC) classification by adapting self-learning hierarchical results. This study evaluated the classification performance of HLEL, random forest (RF), support vector machine (SVM), artificial neural network (ANN), and Gaussian Naive Bayes' (GNB) with GF-6 WFV multispectral images as the database and Sanjiang National Nature Reserve I in China as the study area. The results showed that the HLEL algorithm has an advantage over a single algorithm when using the feature set filtered by ReliefF, even if the complete feature set is used without feature selection. The overall accuracy of the classification results is improved by 0.4%, 5.8%, 1.3%, and 9.0% when compared to the four single algorithms RF, SVM, ANN, and GNB, respectively. In robustness tests, HLEL consistently outperformed the other algorithms for different training sample set sizes, with overall accuracy improvements of 0.4–6.9%, 5.8–11.5%, 1.3–4.4%, and 9.0–14.0% compared with RF, SVM, ANN, and GNB, respectively. The overall accuracy standard deviation of HLEL is 1.22%, proving that HLEL has low sensitivity to the size of the training data. The HLEL algorithm demonstrates exceptional precision when applied to diverse datasets. The overall accuracy of HLEL in Sentinel 2, Landsat 8, and GF-6 WFV increased by 3.1% to 22.7%, 3.1% to 39.0%, and 0.4% to 9.0%, respectively, when compared to the overallaccuracy of the basicclassifiers. The results demonstrate the feasibility of HLEL, an algorithm with high accuracy and data robustness, in applying wetland LULC classification and providing a new solution for wetland cover mapping.

Real World Data Studies of Antineoplastic Drugs: How Can They Be Improved to Steer Everyday Use in the Clinic?

There is a growing interest in real world evidence when developing antineoplastic drugs owing to the shorter length of time and low costs compared to randomised controlled trials. External validity of studies in the regulatory phase can be enhanced by complementing randomised controlled trials with real world evidence. Furthermore, the use of real world evidence ensures the inclusion of patients often excluded from randomised controlled trials such as the elderly, certain ethnicities or those from certain geographical areas. This review explores approaches in which real world data may be integrated with randomised controlled trials. One approach is by using big data, especially when investigating drugs in the antineoplastic setting. This can even inform artificial intelligence thus ensuring faster and more precise diagnosis and treatment decisions. Pragmatic trials also offer an approach to examine the effectiveness of novel antineoplastic drugs without evading the benefits of randomised controlled trials. A well-designed pragmatic trial would yield results with high external validity by employing a simple study design with a large sample size and diverse settings. Although randomised controlled trials can determine efficacy of antineoplastic drugs, effectiveness in the real world may differ. The need for pragmatic trials to help guide healthcare decision-making led to the development of trials within cohorts (TWICs). TWICs make use of cohorts to conduct multiple randomised controlled trials while maintaining characteristics of real world data in routine clinical practice. Although real world data is often affected by incomplete data and biases such as selection and unmeasured biases, the use of big data and pragmatic approaches can improve the use of real world data in the development of antineoplastic drugs that can in turn steer decision-making in clinical practice.

A Large Benchmark Dataset for Individual Sheep Face Recognition

The mutton sheep breeding industry has transformed significantly in recent years, from traditional grassland free-range farming to a more intelligent approach. As a result, automated sheep face recognition systems have become vital to modern breeding practices and have gradually replaced ear tagging and other manual tracking techniques. Although sheep face datasets have been introduced in previous studies, they have often involved pose or background restrictions (e.g., fixing of the subject’s head, cleaning of the face), which restrict data collection and have limited the size of available sample sets. As a result, a comprehensive benchmark designed exclusively for the evaluation of individual sheep recognition algorithms is lacking. To address this issue, this study developed a large-scale benchmark dataset, Sheepface-107, comprising 5350 images acquired from 107 different subjects. Images were collected from each sheep at multiple angles, including front and back views, in a diverse collection that provides a more comprehensive representation of facial features. In addition to the dataset, an assessment protocol was developed by applying multiple evaluation metrics to the results produced by three different deep learning models: VGG16, GoogLeNet, and ResNet50, which achieved F1-scores of 83.79%, 89.11%, and 93.44%, respectively. A statistical analysis of each algorithm suggested that accuracy and the number of parameters were the most informative metrics for use in evaluating recognition performance.

Percutaneous cannulation is associated with lower rate of severe neurological complication in femoro-femoral ECPR: results from the Extracorporeal Life Support Organization Registry

BackgroundPercutaneous cannulation is now accepted as the first-line strategy for extracorporeal cardiopulmonary resuscitation (ECPR) in adults. However, previous studies comparing percutaneous cannulation to surgical cannulation have been limited by small sample size and single-center settings. This study aimed to compare in-hospital outcomes in cardiac arrest (CA) patients who received femoro-femoral ECPR with percutaneous vs surgical cannulation.MethodsAdults with refractory CA treated with percutaneous (percutaneous group) or surgical (surgical group) femoro-femoral ECPR between January 2008 and December 2019 were extracted from the international Extracorporeal Life Support Organization registry. The primary outcome was severe neurological complication. Multivariable logistic regression analyses were performed to assess the association between percutaneous cannulation and in-hospital outcomes.ResultsAmong 3575 patients meeting study inclusion, 2749 (77%) underwent percutaneous cannulation. The proportion of patients undergoing percutaneous cannulation increased from 18% to 89% over the study period (p < 0.001 for trend). Severe neurological complication (13% vs 19%; p < 0.001) occurred less frequently in the percutaneous group compared to the surgical group. In adjusted analyses, percutaneous cannulation was independently associated with lower rate of severe neurological complication (odds ratio [OR] 0.62; 95% CI 0.46–0.83; p = 0.002), similar rates of in-hospital mortality (OR 0.93; 95% CI 0.73–1.17; p = 0.522), limb ischemia (OR 0.84; 95% CI 0.58–1.20; p = 0.341) and cannulation site bleeding (OR 0.90; 95% CI 0.66–1.22; p = 0.471). The comparison of outcomes provided similar results across different levels of center percutaneous experience or center ECPR volume.ConclusionsAmong adults receiving ECPR, percutaneous cannulation was associated with probable lower rate of severe neurological complication, and similar rates of in-hospital mortality, limb ischemia and cannulation site bleeding.

Global incidence of non-alcoholic fatty liver disease: A systematic review and meta-analysis of 63 studies and 1,201,807 persons

The prevalence of non-alcoholic fatty liver disease (NAFLD) is increasing. We aimed to estimate the pooled global NAFLD incidence. We performed a systematic review and meta-analysis of cohort studies of adults without NAFLD at baseline to evaluate the global incidence of ultrasound-diagnosed NAFLD. A total of 63 eligible studies (1,201,807 persons) were analyzed. Studies were from Mainland China/Hong Kong (n= 26), South Korea (n= 22), Japan (n= 14), other (n= 2, Sri Lanka, Israel); 63.8% were clinical center studies; median study year 2000 to 2016; 87% were good quality. Among the 1,201,807 persons at risk, 242,568 persons developed NAFLD, with an incidence rate of 4,612.8 (95% CI 3,931.5-5,294.2) per 100,000 person-years and no statistically significant differences by study sample size (p=0.90) or study setting (p=0.055). Males had higher incidence vs. females (5,943.8 vs. 3,671.7, p=0.0013). Both the obese (vs. non-obese) and the overweight/obese groups (vs. normal weight) were about threefold more likely to develop NAFLD (8,669.6 vs. 2,963.9 and 8,416.6 vs. 3,358.2, respectively) (both p <0.0001). Smokers had higher incidence than non-smokers (8,043.2 vs. 4,689.7, p=0.046). By meta-regression, adjusting for study year, study setting, and study location, study period of 2010 or after and study setting were associated with increased incidence (p=0.010 and p=0.055, respectively). By country, China had a higher NAFLD incidence compared to non-China regions (p=0.012) and Japan a lower incidence compared to non-Japan regions (p=0.005). NAFLD incidence is increasing with a current estimate of 4,613 new cases per 100,000 person-years. Males and overweight/obese individuals had significantly higher incidence rates compared to females and those of normal weight. Public health interventions for prevention of NAFLD are needed with a special emphasis on males, overweight/obese individuals, and higher risk regions. Non-alcoholic fatty liver disease (NAFLD) affects approximately 30% of people worldwide and appears to be increasing, but data to estimate the incidence rate are limited. In this meta-analytic study of over 1.2 million people, we estimated an incidence rate of NAFLD of 46.13 per 1,000 person-years with significant differences by sex, BMI, geography, and time-period. As treatment options for NAFLD remain limited, prevention of NAFLD should remain the focus of public health strategies. Studies such as these can help policy makers in determining which and whether their interventions are impactful.

Pricing and sample set strategies of data providers under quality information asymmetry

Current data transactions are usually driven by data providers, and data buyers have little information about the data they will buy. In turn, this information asymmetry impacts the data providers’ sales decisions. In this paper, we consider a data provider with private quality information to sell a data set to multiple potential data buyers, and the data provider signals quality information to the potential buyers by setting the price and sample set size. We study how the data provider decides the optimal trading strategy in terms of price and sample set size based on different characteristics of data buyers through a signalling game. It is found that the degree of information asymmetry and the cost of providing a sample set affect the data provider’s choice of whether to signal data quality truthfully or to hide quality information. The joint signal of price and sample set is a better signalling option for data providers to achieve higher profits than the price signal alone. Tracing data trading back to the production process of datasets, the increased information asymmetry and inefficient data production make data providers more inclined to generate low-quality datasets, which adversely affects the data trading market.

Data Augmentation in High Dimensional Low Sample Size Setting Using a Geometry-Based Variational Autoencoder.

In this paper, we propose a new method to perform data augmentation in a reliable way in the High Dimensional Low Sample Size (HDLSS) setting using a geometry-based variational autoencoder (VAE). Our approach combines the proposal of 1) a new VAE model, the latent space of which is modeled as a Riemannian manifold and which combines both Riemannian metric learning and normalizing flows and 2) a new generation scheme which produces more meaningful samples especially in the context of small data sets. The method is tested through a wide experimental study where its robustness to data sets, classifiers and training samples size is stressed. It is also validated on a medical imaging classification task on the challenging ADNI database where a small number of 3D brain magnetic resonance images (MRIs) are considered and augmented using the proposed VAE framework. In each case, the proposed method allows for a significant and reliable gain in the classification metrics. For instance, balanced accuracy jumps from 66.3% to 74.3% for a state-of-the-art convolutional neural network classifier trained with 50 MRIs of cognitively normal (CN) and 50 Alzheimer disease (AD) patients and from 77.7% to 86.3% when trained with 243 CN and 210 AD while improving greatly sensitivity and specificity metrics.

Assessment of suppliers’ collaboration on construction project in Ebonyi state

In Nigeria, the concept of suppliers’ collaboration is still a new area to construction stakeholders. The construction companies in Ebonyi State are faced with enormous challenges which involve the integration of all construction processes and activities of different agents and specialists hired by clients for successful project delivery. To address the problem of construction project in Ebonyi State, construction stakeholders require collaboration which suppliers’ collaboration is the most pertinent. The aim is to assess the factors affecting contractors–suppliers relationship on successful construction project delivery in Ebonyi State. The general objective is to identify and determine the factors affecting contractors–suppliers relationship in construction industry and impact of supplier’s collaboration on the cost of construction project in Ebonyi State. The research sought to ensure at least 95% level of confidence with a total population of 400 and sample size set to 250. Primary and secondary sources were used to generate data with the aid of structured questionnaire and oral interviews directed to both private firms and public firms in construction industry. In total, 250 forms containing a set of question were submitted randomly to selected medium and large-sized construction firms in Ebonyi State. Data collected were presented in tables containing frequencies of the responses and their corresponding percentages. Frequency distribution, relative importance index (R11), Chi-square (X2), Python spyder IDE, statistical package for social science students(SPSS) and Pearson product moment correlation coefficient were the analytical tools used in the study. The result of the analyses revealed that, correlation matrix used to measure the relationship between contractors and supplier variables. The matrix depicts there is a linear correlation between all possible pairs about contractors and suppliers. There exist a relationship between contractors and suppliers as shown with other pairs in the correlation matrix that connects to objective number 3 and addresses the research question number 3. This implies that as the contractor-supplier approach short term or/and long term increases, the supplier collaboration on the cost of construction project in the Ebonyi State is expected to increase. The study concludes that the main drivers of the success of supplier collaboration appear to be the elements of trust and teamwork. It is these which ensure the continued relationship and ensure that both parties derive the benefits necessary for the partnership to continue and therefore; made the following recommendation. A detailed schedule of materials supply should be provided by the contractor. The schedule should contain the time required for material to arrive on site. Thus, results in reducing unproductive time on site and cost saving.

Sample size requirements for riverbank macrolitter characterization

Anthropogenic litter is omnipresent in terrestrial and freshwater systems, and can have major economic and ecological impacts. Monitoring and modeling of anthropogenic litter comes with large uncertainties due to the wide variety of litter characteristics, including size, mass, and item type. It is unclear as to what the effect of sample set size is on the reliability and representativeness of litter item statistics. Reliable item statistics are needed to (1) improve monitoring strategies, (2) parameterize litter in transport models, and (3) convert litter counts to mass for stock and flux calculations. In this paper, we quantify sample set size requirement for riverbank litter characterization, using a database of more than 14,000 macrolitter items (&amp;gt;0.5 cm), sampled for 1 year at eight riverbank locations along the Dutch Rhine, IJssel, and Meuse rivers. We use this database to perform a Monte Carlo based bootstrap analysis on the item statistics, to determine the relation between sample size and variability in the mean and median values. Based on this, we present sample set size requirements, corresponding to selected uncertainty and confidence levels. Optima between sampling effort and information gain is suggested (depending on the acceptable uncertainty level), which is a function of litter type heterogeneity. We found that the heterogeneity of the characteristics of litter items varies between different litter categories, and demonstrate that the minimum required sample set size depends on the heterogeneity of the litter category. This implies that more items of heterogeneous litter categories need to be sampled than of heterogeneous item categories to reach the same uncertainty level in item statistics. For example, to describe the mean mass the heterogeneous category soft fragments (&amp;gt;2.5 cm) with 90% confidence, 990 items were needed, while only 39 items were needed for the uniform category metal bottle caps. Finally, we use the heterogeneity within litter categories to assess the sample size requirements for each river system. All data collected for this study are freely available, and may form the basis of an open access global database which can be used by scientists, practitioners, and policymakers to improve future monitoring strategies and modeling efforts.