Physical Sciences and Mathematics Commons^™

The Efficiency And Accuracy Of Yolo For Neonate Face Detection In The Clinical Setting, Jacqueline Hausmann

USF Tampa Graduate Theses and Dissertations

There are many face detection classification models available for download and use in the modern technological world. Based in the field of deep neural networks, these off-the-shelf solutions are generally inadequate to solve real world challenges. This work presents how current approaches biased towards detecting adult human faces must be modified in order to better accommodate face detection of the neonate in a NICU setting.

YOLO is a powerful object detection algorithm. Due to optimizations such as Cross mini-batch Normalization, Modified Spatial Attention Modules, Modified Path Aggregation Networks, Self-Adversarial Training, Mosaic Data Augmentation, DropBox Regularization, Multi-Input Weighted Residual Connections and …

Comparing Variable Importance In Prediction Of Silence Behaviours Between Random Forest And Conditional Inference Forest Models., Stephen Barrett Dr, Geraldine Gray Dr, Colm Mcguinness Dr, Michael Knoll Dr.

Articles

This paper explores variable importance metrics of Conditional Inference Trees (CIT) and classical Classification And Regression Trees (CART) based Random Forests. The paper compares both algorithms variable importance rankings and highlights why CIT should be used when dealing with data with different levels of aggregation. The models analysed explored the role of cultural factors at individual and societal level when predicting Organisational Silence behaviours.

Reducing Age Bias In Machine Learning: An Algorithmic Approach, Adriana Solange Garcia De Alford, Steven K. Hayden, Nicole Wittlin, Amy Atwood

SMU Data Science Review

In this paper, we study the prevalence of bias in machine learning; we explore the life cycle phases where bias is potentially introduced into a machine learning model; and lastly, we present how adversarial learning can be leveraged to measure unwanted bias and unfair behavior from a machine learning algorithm. This study focuses particularly on the topics of age bias in predicting employee attrition and presents a practical approach for how adversarial learning can be successful in mitigating age bias. To measure bias, we calculate group fairness metrics across five-year age groups and evaluate fairness between a baseline predictive model …

A New Exponential Approach For Reducing The Mean Squared Errors Of The Estimators Of Population Mean Using Conventional And Non-Conventional Location Parameters, Housila P. Singh, Anita Yadav

Journal of Modern Applied Statistical Methods

Classes of ratio-type estimators t (say) and ratio-type exponential estimators t_e (say) of the population mean are proposed, and their biases and mean squared errors under large sample approximation are presented. It is the class of ratio-type exponential estimators t_e provides estimators more efficient than the ratio-type estimators.

An Application Of Machine Learning To Explore Relationships Between Factors Of Organisational Silence And Culture, With Specific Focus On Predicting Silence Behaviours, Stephen Barrett Dr

Articles

Research indicates that there are many individual reasons why people do not speak up when confronted with situations that may concern them within their working environment. One of the areas that requires more focused research is the role culture plays in why a person may remain silent when such situations arise. The purpose of this study is to use data science techniques to explore the patterns in a data set that would lead a person to engage in organisational silence. The main research question the thesis asks is: Is Machine Learning a tool that Social Scientists can use with respect …

Transect Survey Biases And Correction Methods In Southern Africa, Erika Swenson, Larkin Powell

UCARE Research Products

In southern Africa, transect surveys and distance-based analyses are often used to obtain density and population estimates for species in large reserves or management zones. However, these estimates may be biased by unnaturally large concentrations of animals at waterholes that are on or near the path of the transect. We used empirical survey data from the Namibrand Nature Reserve in southwest Namibia to parameterize spatial simulations in which we distributed gemsbok (Oryx gazella) on a grid along a 50-kilometer transect. We created multiple simulations with and without waterholes to determine how the proportion of animals clumped at the …

The Importance Of Type I Error Rates When Studying Bias In Monte Carlo Studies In Statistics, Michael Harwell

Journal of Modern Applied Statistical Methods

Two common outcomes of Monte Carlo studies in statistics are bias and Type I error rate. Several versions of bias statistics exist but all employ arbitrary cutoffs for deciding when bias is ignorable or non-ignorable. This article argues Type I error rates should be used when assessing bias.

Exploring Composite Dataset Biases For Heart Sound Classification, Davoud Shariat Panah, Andrew Hines, Susan Mckeever

Conference papers

In the last few years, the automatic classification of heart sounds has been widely studied as a screening method for heart disease. Some of these studies have achieved high accuracies in heart abnormality prediction. However, for such models to assist clinicians in the detection of heart abnormalities, it is of critical importance that they are generalisable, working on unseen real-world data. Despite the importance of generalisability, the presence of bias in the leading heart sound datasets used in these studies has remained unexplored. In this paper, we explore the presence of potential bias in heart sound datasets. Using a small …

Implementation Considerations For Mitigating Bias In Supervised Machine Learning, Bardia Bijani Aval

CSB and SJU Distinguished Thesis

Machine Learning (ML) is an important component of computer science and a mainstream way of making sense of large amounts of data. Although the technology is establishing new possibilities in different fields, there are also problems to consider, one of which is bias. Due to the inductive reasoning of ML algorithms in creating mathematical models, the predictions and trends found by the models will never necessarily be true – just more or less probable. Knowing this, it is unreasonable for us to expect the applied deductive reasoning of these models to ever be fully unbiased. Therefore, it is important that …

Investigating Bias In Facial Analysis Systems: A Systematic Review, Ashraf Khalil, Soha Glal Ahmed, Asad Masood Khattak, Nabeel Al-Qirim

All Works

© 2013 IEEE. Recent studies have demonstrated that most commercial facial analysis systems are biased against certain categories of race, ethnicity, culture, age and gender. The bias can be traced in some cases to the algorithms used and in other cases to insufficient training of algorithms, while in still other cases bias can be traced to insufficient databases. To date, no comprehensive literature review exists which systematically investigates bias and discrimination in the currently available facial analysis software. To address the gap, this study conducts a systematic literature review (SLR) in which the context of facial analysis system bias is …

Efficient Class Of Estimators For Finite Population Mean Using Auxiliary Information In Two-Occasion Successive Sampling, G. N. Singh, Mohd Khalid

Journal of Modern Applied Statistical Methods

In the case of sampling on two occasions, a class of estimators is considered which uses information on the first occasion as well as the second occasion in order to estimate the population means on the current (second) occasion. The usefulness of auxiliary information in enhancing the efficiency of this estimation is examined through the class of proposed estimators. Some properties of the class of estimators and a strategy of optimum replacement are discussed. The proposed class of estimators were empirically compared with the sample mean estimator in the case of no matching. The established optimum estimator, which is a …

A Strategy For Using Bias And Rmse As Outcomes In Monte Carlo Studies In Statistics, Michael Harwell

Journal of Modern Applied Statistical Methods

To help ensure important patterns of bias and accuracy are detected in Monte Carlo studies in statistics this paper proposes conditioning bias and root mean square error (RMSE) measures on estimated Type I and Type II error rates. A small Monte Carlo study is used to illustrate this argument.

Fairness And Discrimination In Recommendation And Retrieval, Michael D. Ekstrand, Robin Burke, Fernando Diaz

Computer Science Faculty Publications and Presentations

Fairness and related concerns have become of increasing importance in a variety of AI and machine learning contexts. They are also highly relevant to recommender systems and related problems such as information retrieval, as evidenced by the growing literature in RecSys, FAT*, SIGIR, and special sessions such as the FATREC and FACTS-IR workshops and the Fairness track at TREC 2019; however, translating algorithmic fairness constructs from classification, scoring, and even many ranking settings into recommendation and other information access scenarios is not a straightforward task. This tutorial will help orient RecSys researchers to algorithmic fairness, understand how concepts do and …

Exploring The Impact Of (Not) Changing Default Settings In Algorithmic Crime Mapping - A Case Study Of Milwaukee, Wisconsin, Md Romael Haque, Katy Weathington, Shion Guha

Computer Science Faculty Research and Publications

Policing decisions, allocations and outcomes are determined by mapping historical crime data geo-spatially using popular algorithms. In this extended abstract, we present early results from a mixed-methods study of the practices, policies, and perceptions of algorithmic crime mapping in the city of Milwaukee, Wisconsin. We investigate this differential by visualizing potential demographic biases from publicly available crime data over 12 years (2005-2016) and conducting semi-structured interviews of 19 city stakeholders and provide future research directions from this study.

A Comparison Of Bayesian Estimation Techniques In A Multidimensional Two-Parameter Partial Credit Item Response Model, Peiyan Liu

Electronic Theses and Dissertations

Bayesian estimation methods have shown better performance than the traditional Marginal Maximum Likelihood (MML) estimation method for parameter estimation in relatively simple item response models. However, extant literature is lacking on the investigation of Bayesian parameter estimation approaches for a multidimensional two parameter partial credit (M2PPC) model, therefore this simulation study investigated the performance of two Bayesian Markov Chain Monte Carlo (MCMC) algorithms: Gibbs Sampler and Hamiltonian Monte Carlo-No-U-Turn-Sampler (HMC-NUTS) for M2PPC models' parameter estimation. It compared the estimation accuracy and computing speed in different combinations of situations, including prior choices, test lengths, and the relationships between dimensions.

The datasets …

Emerging Roles Of Virtual Patients In The Age Of Ai, C. Donald Combs, P. Ford Combs

Computational Modeling & Simulation Engineering Faculty Publications

Today's web-enabled and virtual approach to medical education is different from the 20th century's Flexner-dominated approach. Now, lectures get less emphasis and more emphasis is placed on learning via early clinical exposure, standardized patients, and other simulations. This article reviews literature on virtual patients (VPs) and their underlying virtual reality technology, examines VPs' potential through the example of psychiatric intake teaching, and identifies promises and perils posed by VP use in medical education.

Controlling For Confounding Via Propensity Score Methods Can Result In Biased Estimation Of The Conditional Auc: A Simulation Study, Hadiza I. Galadima, Donna K. Mcclish

Community & Environmental Health Faculty Publications

In the medical literature, there has been an increased interest in evaluating association between exposure and outcomes using nonrandomized observational studies. However, because assignments to exposure are not random in observational studies, comparisons of outcomes between exposed and nonexposed subjects must account for the effect of confounders. Propensity score methods have been widely used to control for confounding, when estimating exposure effect. Previous studies have shown that conditioning on the propensity score results in biased estimation of conditional odds ratio and hazard ratio. However, research is lacking on the performance of propensity score methods for covariate adjustment when estimating the …

Examining Patterns In Nest Predation Using Artificial Nests, Victoria L. Simonsen

School of Natural Resources: Dissertations, Theses, and Student Research

The use of artificial nests to study the predation of avian nests has faced disregard by ecologists due to inconsistencies found between the survival rates of real and artificial nests across studies and reviews. The negative perception of artificial nests providing an inconsistent assessment of survival has thus fostered the perception that artificial nests are a secondary option to be used to overcome logistical hurdles associated with achieving sufficient sample sizes in systems where study species are rare or elusive, or as merely a preliminary method to study predation across gradients. We argue that the greatest mistake ecologists have made …

Evaluation Of Using The Bootstrap Procedure To Estimate The Population Variance, Nghia Trong Nguyen

Electronic Theses and Dissertations

The bootstrap procedure is widely used in nonparametric statistics to generate an empirical sampling distribution from a given sample data set for a statistic of interest. Generally, the results are good for location parameters such as population mean, median, and even for estimating a population correlation. However, the results for a population variance, which is a spread parameter, are not as good due to the resampling nature of the bootstrap method. Bootstrap samples are constructed using sampling with replacement; consequently, groups of observations with zero variance manifest in these samples. As a result, a bootstrap variance estimator will carry a …

Investigation Of Model Stacking For Drug Sensitivity Prediction, Kevin Matlock, Carlos De Niz, Raziur Rahman, Souparno Ghosh, Ranadip Pal

Department of Statistics: Faculty Publications

Background: A significant problem in precision medicine is the prediction of drug sensitivity for individual cancer cell lines. Predictive models such as Random Forests have shown promising performance while predicting from individual genomic features such as gene expressions. However, accessibility of various other forms of data types including information on multiple tested drugs necessitates the examination of designing predictive models incorporating the various data types.

Results: We explore the predictive performance of model stacking and the effect of stacking on the predictive bias and squarred error. In addition we discuss the analytical underpinnings supporting the advantages of stacking in reducing …

Exploring Author Gender In Book Rating And Recommendation, Michael D. Ekstrand, Mucun Tian, Mohammed R. Imran Kazi, Hoda Mehrpouyan, Daniel Kluver

Computer Science Faculty Publications and Presentations

Collaborative filtering algorithms find useful patterns in rating and consumption data and exploit these patterns to guide users to good items. Many of the patterns in rating datasets reflect important real-world differences between the various users and items in the data; other patterns may be irrelevant or possibly undesirable for social or ethical reasons, particularly if they reflect undesired discrimination, such as gender or ethnic discrimination in publishing. In this work, we examine the response of collaborative filtering recommender algorithms to the distribution of their input data with respect to a dimension of social concern, namely content creator gender. Using …

Evolution Of Bias In Human And Machine Learning Algorithm Interaction, Wenlong Sun, Olfa Nasraoui, Patrick Shafto

Commonwealth Computational Summit

Human algorithm interaction:

• people are now affected by the output of all types of machine learning algorithms.
• social media, blogs, social networks, and other services and applications.

Motivation:

• ML algorithm relied on reliable labels from experts to build prediction.
• However, ML algorithm started to receive data from the more general population.
• The interaction leads to biased result which is caused by ingesting unchecked information from general population, such as biased samples and biased labels.

Impact Of Home Visit Capacity On Genetic Association Studies Of Late-Onset Alzheimer's Disease, David W. Fardo, Laura E. Gibbons, Shubhabrata Mukherjee, M. Maria Glymour, Wayne Mccormick, Susan M. Mccurry, James D. Bowen, Eric B. Larson, Paul K. Crane

Biostatistics Faculty Publications

INTRODUCTION—Findings for genetic correlates of late-onset Alzheimer's disease (LOAD) in studies that rely solely on clinic visits may differ from those with capacity to follow participants unable to attend clinic visits.

METHODS—We evaluated previously identified LOAD-risk single nucleotide variants in the prospective Adult Changes in Thought study, comparing hazard ratios (HRs) estimated using the full data set of both in-home and clinic visits (n = 1697) to HRs estimated using only data that were obtained from clinic visits (n = 1308). Models were adjusted for age, sex, principal components to account for ancestry, and additional health indicators.

RESULTS …

Intuition: Role, Biases, Cognitive Basis, And A Hypothetical Synergistic Explanation Of Intuitive Brain Operations, Jens G. Pohl

Collaborative Agent Design (CAD) Research Center

This paper explores the characteristics of the intuitive responses that are generated by our brain continuously in an automatic and effortless manner. However, while intuition is a very powerful mechanism, it is also subject to many biasing influences. The author discusses the role of intuition, examines representative examples of biasing influences, compares cognitive theories of intuition advanced by Simon (2002), Klein (2003 and 1999), and Kahneman (2011), and then advances a hypothetical explanation of the neurological operations underlying intuition based on Hebbian rules (Hebb 1949) of plasticity in combination with synergetic principles.

Exponential Chain Dual To Ratio Cum Dual To Product Estimator For Finite Population Mean In Double Sampling Scheme, Yater Tato, B. K. Singh

Applications and Applied Mathematics: An International Journal (AAM)

This paper considers an exponential chain dual to ratio cum dual to product estimator for estimating finite population mean using two auxiliary variables in double sampling scheme when the information on another additional auxiliary variable is available along with the main auxiliary variable. The expressions for bias and mean square error of the asymptotically optimum estimator are identified in two different cases. The optimum value of the first phase and second phase sample size has been obtained for the fixed cost of survey. To illustrate the results, theoretical and empirical studies have also been carried out to judge the merits …

Effective Estimation Strategy Of Finite Population Variance Using Multi-Auxiliary Variables In Double Sampling, Reba Maji, G. N. Singh, Arnab Bandyopadhyay

Journal of Modern Applied Statistical Methods

Estimation of population variance in two-phase (double) sampling is considered using information on multiple auxiliary variables. An unbiased estimator is proposed and its properties are studied under two different structures. The superiority of the suggested estimator over some contemporary estimators of population variance was established through empirical studies from a natural and an artificially generated dataset.

Gender Bias In It Hiring Practices: An Ethical Analysis, Harmony L. Alford

Student Scholarship – Computer Science

With the current movement to increase the number of women in STEM-related careers, modified IT hiring practices may be considered debatably unethical. Studies cited in this work have asserted that female representation in STEM fields is integral not only to encouraging continued progression toward gender equality in the workplace but also to creating more inclusive products. In turn, some argue that when faced with reasonably comparable female and male candidates, a hiring manager should select the female candidate in order to increase the female representation in the company and provide a female perspective. However, it is simultaneously debatably unethical and …

Efficient And Unbiased Estimation Procedure Of Population Mean In Two-Phase Sampling, Reba Maji, Arnab Bandyopadhyay, G. N. Singh

Journal of Modern Applied Statistical Methods

In this paper, an unbiased regression-ratio type estimator has been developed for estimating the population mean using two auxiliary variables in double sampling. Its properties are studied under two different cases. Empirical studies and graphical simulation have been done to demonstrate the efficiency of the proposed estimator over other estimators.

An Improved Generalized Estimation Procedure Of Current Population Mean In Two-Occasion Successive Sampling, G. N. Singh, Alok Kumar Singh, Anup Kumar Sharma

Journal of Modern Applied Statistical Methods

The present work is an attempt to make use of several auxiliary variables on both occasions for improving the precision of estimates for the current population mean in two-occasion successive sampling. A generalized exponential-cum-regression type estimator of the current population mean is proposed and its optimum replacement strategy has been discussed. Empirical studies are carried out to show the dominance of the proposed estimation procedure over the sample mean estimator and natural successive sampling estimator. Empirical results have been interpreted and suitable recommendations are put forward to survey practitioners.

Estimation Of P(X > Y) When X And Y Are Dependent Random Variables Using Different Bivariate Sampling Schemes, Hani M. Samawi, Amal Helu, Haresh Rochani, Jingjing Yin, Daniel Linder

Biostatistics Faculty Publications

The stress-strength models have been intensively investigated in the literature in regards of estimating the reliability θ = P (X > Y) using parametric and nonparametric approaches under different sampling schemes when X and Y are independent random variables. In this paper, we consider the problem of estimating θ when (X, Y) are dependent random variables with a bivariate underlying distribution. The empirical and kernel estimates of θ = P (X > Y), based on bivariate ranked set sampling (BVRSS) are considered, when (X, Y) are paired dependent continuous random variables. The estimators obtained are compared to their counterpart, bivariate simple random …