Context Of Machine Learning Research Articles

Forecasting Psychological Distress Across Waves Of The COVID-19 Pandemic: A Multi-country Longitudinal Study

Abstract The current study used a longitudinal multi-country cohort survey (N = 8013 at baseline; age (M) = 40.3, SD = 15; female 76%) to investigate whether psychological distress assessed at five waves (May-June 2020; September-October 2020; December 2020-January 2021; March-April 2021 and March-May 2022) during the COVID-19 outbreak can be forecasted by a variety of socio-demographic and psychological- and COVID-related factors assessed at baseline. Psychological distress was defined by elevated depression and/or anxiety scores (PHQ-9 and GAD-7 > = 10). Predictors included demographic characteristics (e.g., gender, age, education), infections with COVID-19, loss of income, substance abuse, domestic violence, contamination fear, basic value orientations, social support and coping strategies. A Random Forest (RF) prediction model within a machine learning context was used to estimate the association between the independent variables and the presence of psychological distress among respondents completing all waves (N = 1052). Preliminary RF results showed that psychological distress at multiple waves, both during the initial phases (2020) as well as the more protracted phases (2021 and 2022), was adequately forecasted by scores on a number of basic value orientations (e.g., lower tradition and conformity, higher security and predictability), by demographic characteristics (e.g., younger age), by the fear of being contaminated, and by lower levels of social support and higher levels of loneliness. It is concluded that during a pandemic, certain personal values may elicit more vulnerability for psychological distress than others. Effective psychological intervention strategies should encompass personal traits and values and factors such as low social support levels and feelings of loneliness to mitigate distress. These insights can help in accurately pinpointing high-risk groups for heightened distress in future public health crises.

Advancing AI Data Ethics in Nursing: Future Directions for Nursing Practice, Research, and Education.

The ethics of artificial intelligence (AI) are increasingly recognized due to concerns such as algorithmic bias, opacity, trust issues, data security, and fairness. Specifically, machine learning algorithms, central to AI technologies, are essential in striving for ethically sound systems that mimic human intelligence. These technologies rely heavily on data, which often remain obscured within complex systems and must be prioritized for ethical collection, processing, and usage. The significance of data ethics in achieving responsible AI was first highlighted in the broader context of health care and subsequently in nursing. This viewpoint explores the principles of data ethics, drawing on relevant frameworks and strategies identified through a formal literature review. These principles apply to real-world and synthetic data in AI and machine-learning contexts. Additionally, the data-centric AI paradigm is briefly examined, emphasizing its focus on data quality and the ethical development of AI solutions that integrate human-centered domain expertise. The ethical considerations specific to nursing are addressed, including 4 recommendations for future directions in nursing practice, research, and education and 2 hypothetical nurse-focused ethical case studies. The primary objectives are to position nurses to actively participate in AI and data ethics, thereby contributing to creating high-quality and relevant data for machine learning applications.

MAPLES-DR: MESSIDOR Anatomical and Pathological Labels for Explainable Screening of Diabetic Retinopathy

Reliable automatic diagnosis of Diabetic Retinopathy (DR) and Macular Edema (ME) is an invaluable asset in improving the rate of monitored patients among at-risk populations and in enabling earlier treatments before the pathology progresses and threatens vision. However, the explainability of screening models is still an open question, and specifically designed datasets are required to support the research. We present MAPLES-DR (MESSIDOR Anatomical and Pathological Labels for Explainable Screening of Diabetic Retinopathy), which contains, for 198 images of the MESSIDOR public fundus dataset, new diagnoses for DR and ME as well as new pixel-wise segmentation maps for 10 anatomical and pathological biomarkers related to DR. This paper documents the design choices and the annotation procedure that produced MAPLES-DR, discusses the interobserver variability and the overall quality of the annotations, and provides guidelines on using the dataset in a machine learning context.

Enhancing Intrusion Detection in Wireless Sensor Networks through Machine Learning Techniques and Context Awareness Integration

Enhancing Intrusion Detection in Wireless Sensor Networks through Machine Learning Techniques and Context Awareness Integration

Adaptive Method for Feature Selection in the Machine Learning Context

Feature selection is a fundamental aspect of machine learning that is crucial for improving the accuracy and efficiency of models. It carefully analyzes the abundance of data to identify the most significant characteristics, hence improving the accuracy of predictions and minimizing the likelihood of model overfitting. This technique not only optimizes model training by reducing computational requirements, but also enhances the model's interpretability, resulting in more transparent and reliable predictions. The deliberate omission of unnecessary variables is a process of improving the model and also constitutes a crucial measure toward achieving more flexible and comprehensible results in machine learning. An analysis to assess the effectiveness of feature selection on regression models was conducted. The impact was measured using Mean Squared Error (MSE) metrics. A variety of regression algorithms were evaluated, and then feature selection techniques, including statistical and algorithmic methods, such as SelectKBest, PCA, and RFE with Linear Regression and Random Forest, were applied. After selecting the features, linear models demonstrated improvements in mean squared error (MSE), highlighting the value of removing unnecessary data. This study emphasizes the subtle impact of feature selection on model performance, calling for a tailored strategy to maximize prediction accuracy.

Do We Really Need Imputation in AutoML Predictive Modeling?

Numerous real-world data contain missing values, while in contrast, most Machine Learning (ML) algorithms assume complete datasets. For this reason, several imputation algorithms have been proposed to predict and fill in the missing values. Given the advances in predictive modeling algorithms tuned in an Automated Machine Learning context (AutoML) setting, a question that naturally arises is to what extent sophisticated imputation algorithms (e.g., Neural Network based) are really needed, or we can obtain a descent performance using simple methods like Mean/Mode (MM). In this article, we experimentally compare six state-of-the-art representatives of different imputation algorithmic families from an AutoML predictive modeling perspective, including a feature selection step and combined algorithm and hyper-parameter selection. We used a commercial AutoML tool for our experiments, in which we included the selected imputation methods. Experiments ran on 25 binary classification real-world incomplete datasets with missing values and 10 binary classification complete datasets in which synthetic missing values are introduced according to different missingness mechanisms, at varying missing frequencies. The main conclusion drawn from our experiments is that the best method on average is the Denoise AutoEncoder on real-world datasets and the MissForest in simulated datasets, followed closely by MM. In addition, binary indicator variables encoding missingness patterns actually improve predictive performance, on average. Last, although there are cases where Neural-Network-based imputation significantly improves predictive performance, this comes at a great computational cost and requires measuring all feature values to impute new samples.

Feature Imputation using Neutrosophic Set Theory in Machine Learning Regression Context

The prediction context of machine learning aims to discern the underlying patterns that dictate the characteristics to forecast the output. This prediction lacks precision when the input data is not accurate or precise. This study focuses on feature imputation through the application of the neutrosophic set theory. The primary concept involves substituting feature data, which may have accuracy and correctness issues, with neutrosophic variables considering the degrees of truth, indeterminacy, and falsity to produce more precise and resilient predictions. The proposed method was implemented in a specific case study, and the results are analyzed.

Proposal of new models for prediction of the cost of agricultural raw materials in a business intelligence and machine learning context

Proposal of new models for prediction of the cost of agricultural raw materials in a business intelligence and machine learning context

Comparison of dimensionality reduction techniques for the visualisation of chemical space in organometallic catalysis

We have used a Ligand Knowledge Base for bidentate P,P-donor ligands of potential interest to homogeneous catalysis to compare three dimensionality reduction techniques, namely Principal Component Analysis (PCA), Uniform Manifold Approximation and Projection (UMAP) and t-distributed Stochastic Neighbor Embedding (t-SNE). While our previous work on Ligand Knowledge Bases has focused on PCA, here we compare this approach with more recently-published approaches and assess the information retention, visualization, clustering and interpretability which can be achieved for each approach. We find that potential advantages of t-SNE are not realized with a database of the current size (275 entries), and that there is a degree of complementarity between PCA and UMAP. The statistics underlying PCA rely on linear relationships, making interpretation of the resulting plots comparatively straightforward. Since much of chemistry relies on linear structure-property relationships and low-dimensional visualization, the explainability and information retention achieved is attractive. UMAP proved more challenging to interpret, but achieved clear clustering which was often chemically meaningful, and it would be a useful approach for ensuring that distinct subsets of compounds are sampled in a machine-learning context. This analysis also highlighted that the tunability of catalysis achieved through ligand exchange maps well onto some areas of chemical space where closely related ligands cluster, while others represent outliers; these arise from different combinations of steric and electronic effects which chemists will find intuitive.

Universal embedding of autonomous dynamical systems into a Lotka-Volterra-like format

We show that the ordinary differential equations (ODEs) of any deterministic autonomous dynamical system with continuous and bounded rate-field components can be embedded into a quadratic Lotka-Volterra-like form by turning to an augmented set of state variables. The key step consists in expressing the rate equations by employing the Universal Approximation procedure (borrowed from the machine learning context) with logistic sigmoid ‘activation function’. Then, by applying already established methods, the resulting ODEs are first converted into a multivariate polynomial form (also known as generalized Lotka-Volterra), and finally into the quadratic structure. Although the final system of ODEs has a dimension virtually infinite, the feasibility of such a universal embedding opens to speculations and calls for an interpretation at the physical level.

Ultrametrics for context-aware comparison of binary images

Quantitative image comparison has been a key topic in the image processing literature for the past 30 years. The reasons for it are diverse, and so is the range of applications in which measures of comparison are needed. Examples of image processing tasks requiring such measures are the evaluation of algorithmic results (through the comparison of computer-generated results to given ground truth) or the selection of loss/goal functions in a machine learning context. Measures of comparison in literature take different inspirations, and are often tailored to specific needs. Nevertheless, even if some measures of comparison intend to replicate how humans evaluate the similarity of two images, they normally overlook a fundamental characteristic of the way humans perform such evaluation: the context of comparison. In this paper, we present a measure of comparison for binary images that incorporates a sense of context. More specifically, we present a Methodology for the generation of ultrametrics for context-aware comparison of binary images. We test our proposal in the context of boundary image comparison on the BSDS500 benchmark.

Marriage between variable selection and prediction methods to model plant disease risk

Predicting the risk of a disease in a pathosystem based on a set of climatic variables usually requires handling a high number of input variables, many of which are often irrelevant and/or redundant. Building linear predictive models entails not only dimensionality issues but also the negative impact of multicollinearity. Several feature selection methods have proved to be efficient in both linear and non-linear models, regardless of those issues. However, in a machine learning (ML) context, it is necessary to evaluate these feature selection methods embedded into the model fitting algorithm to obtain the greatest accuracy. The aim of this work was to assess different combinations of variable selection methods with linear and non-linear predictors to fit climate-based models that predict the occurrence of a disease in a pathosystem. Four selection methods were compared: stepwise, which is frequently used in linear models, combined with VIF and p-value statistical criteria (Step+VIF+Pv), and other methods commonly used in ML: filter (F), genetic algorithm (GA), and Boruta (B). The disease risk predictors were constructed with a logistic linear regression model (LR) and the random forest (RF) algorithm, using all the available variables and the subgroups of variables selected by each feature selection method. Data from three pathosystems were processed: two involving Begomovirus –one in common bean (Phaseolus vulgaris L) and the other in soybean (Glycine max)– and the third one involving Mal de Rio Cuarto virus in maize (Zea mays L.). The data sets differed in sample size and number of variables. The accuracy of RF prediction did not vary among feature selection methods. Step+VIF+Pv was used to reduce the model outperformed the other feature selection methods in fitting LR. Our proposal suggests that the appropriate pairing of variable selection and prediction models would improve the modeling of plant disease risk.

Fail-tests of DNA methylation clocks, and development of a noise barometer for measuring epigenetic pressure of aging and disease.

This study shows that Elastic Net (EN) DNA methylation (DNAme) clocks have low accuracy of predictions for individuals of the same age and a low resolution between healthy and disease cohorts; caveats inherent in applying linear model to non-linear processes. We found that change in methylation of cytosines with age is, interestingly, not the determinant for their selection into the clocks. Moreover, an EN clock's selected cytosines change when non-clock cytosines are removed from the training data; as expected from optimization in a machine learning (ML) context, but inconsistently with the identification of health markers in a biological context. To address these limitations, we moved from predictions to measurement of biological age, focusing on the cytosines that on average remain invariable in their methylation through lifespan, postulated to be homeostatically vital. We established that dysregulation of such cytosines, measured as the sums of standard deviations of their methylation values, quantifies biological noise, which in our hypothesis is a biomarker of aging and disease. We term this approach a "noise barometer" - the pressure of aging and disease on an organism. These noise-detecting cytosines are particularly important as sums of SD on the entire 450K DNAme array data yield a random pattern through chronology. Testing how many cytosines of the 450K arrays become noisier with age, we found that the paradigm of DNAme noise as a biomarker of aging and disease remarkably manifests in ~1/4 of the total. In that large set even the cytosines that have on average constant methylation through age show increased SDs and can be used as noise detectors of the barometer.

Data re-uploading with a single qudit

Quantum two-level systems, i.e., qubits, form the basis for most quantum machine learning approaches that have been proposed throughout the years. However, higher dimensional quantum systems constitute a promising alternative and are increasingly explored in theory and practice. Here, we explore the capabilities of multi-level quantum systems, so-called qudits, for their use in a quantum machine learning context. We formulate classification and regression problems with the data re-uploading approach and demonstrate that a quantum circuit operating on a single qudit is able to successfully learn highly non-linear decision boundaries of classification problems such as the MNIST digit recognition problem. We demonstrate that the performance strongly depends on the relation between the qudit states representing the labels and the structure of labels in the training data set. Such a bias can lead to substantial performance improvement over qubit-based circuits in cases where the labels, the qudit states, and the operators employed to encode the data are well-aligned. Furthermore, we elucidate the influence of the choice of the elementary operators and show that a squeezing operator is necessary to achieve good performances. We also show that there exists a trade-off for qudit systems between the number of circuit-generating operators in each processing layer and the total number of layers needed to achieve a given accuracy. Finally, we compare classification results from numerically exact simulations and their equivalent implementation on actual IBM quantum hardware. The findings of our work support the notion that qudit-based algorithms exhibit attractive traits and constitute a promising route to increasing the computational capabilities of quantum machine learning approaches.

Federated Edge Intelligence and Edge Caching Mechanisms

Federated learning (FL) has emerged as a promising technique for preserving user privacy and ensuring data security in distributed machine learning contexts, particularly in edge intelligence and edge caching applications. Recognizing the prevalent challenges of imbalanced and noisy data impacting scalability and resilience, our study introduces two innovative algorithms crafted for FL within a peer-to-peer framework. These algorithms aim to enhance performance, especially in decentralized and resource-limited settings. Furthermore, we propose a client-balancing Dirichlet sampling algorithm with probabilistic guarantees to mitigate oversampling issues, optimizing data distribution among clients to achieve more accurate and reliable model training. Within the specifics of our study, we employed 10, 20, and 40 Raspberry Pi devices as clients in a practical FL scenario, simulating real-world conditions. The well-known FedAvg algorithm was implemented, enabling multi-epoch client training before weight integration. Additionally, we examined the influence of real-world dataset noise, culminating in a performance analysis that underscores how our novel methods and research significantly advance robust and efficient FL techniques, thereby enhancing the overall effectiveness of decentralized machine learning applications, including edge intelligence and edge caching.

Protein model quality assessment using rotation-equivariant transformations on point clouds.

Machine learning research concerning protein structure has seen a surge in popularity over the last years with promising advances for basic science and drug discovery. Working with macromolecular structure in a machine learning context requires an adequate numerical representation, and researchers have extensively studied representations such as graphs, discretized 3D grids, and distance maps. As part of CASP14, we explored a new and conceptually simple representation in a blind experiment: atoms as points in 3D, each with associated features. These features-initially just the basic element type of each atom-are updated through a series of neural network layers featuring rotation-equivariant convolutions. Starting from all atoms, we further aggregate information at the level of alpha carbons before making a prediction at the level of the entire protein structure. We find that this approach yields competitive results in protein model quality assessment despite its simplicity and despite the fact that it incorporates minimal prior information and is trained on relatively little data. Its performance and generality are particularly noteworthy in an era where highly complex, customized machine learning methods such as AlphaFold 2 have come to dominate protein structure prediction.

Practical overview of image classification with tensor-network quantum circuits

Circuit design for quantum machine learning remains a formidable challenge. Inspired by the applications of tensor networks across different fields and their novel presence in the classical machine learning context, one proposed method to design variational circuits is to base the circuit architecture on tensor networks. Here, we comprehensively describe tensor-network quantum circuits and how to implement them in simulations. This includes leveraging circuit cutting, a technique used to evaluate circuits with more qubits than those available on current quantum devices. We then illustrate the computational requirements and possible applications by simulating various tensor-network quantum circuits with PennyLane, an open-source python library for differential programming of quantum computers. Finally, we demonstrate how to apply these circuits to increasingly complex image processing tasks, completing this overview of a flexible method to design circuits that can be applied to industrially-relevant machine learning tasks.

Explaining classifiers with measures of statistical association

A new class of probabilistic sensitivity measures that quantifies the degree of association between covariates and generic targets used in classification is proposed, and it is shown that such class possesses the zero-independence property. Corresponding estimators are introduced, asymptotic consistency is proven and bootstrap is used to quantify uncertainty in the estimates. The use of the new dependence measures as explanations in a statistical machine learning context is illustrated. The resulting approach, called Xi-method, is demonstrated through applications involving different data formats: tabular, visual and textual.

Gradient boosting with extreme-value theory for wildfire prediction

This paper details the approach of the team Kohrrelation in the 2021 Extreme Value Analysis data challenge, dealing with the prediction of wildfire counts and sizes over the contiguous US. Our approach uses ideas from extreme-value theory in a machine learning context with theoretically justified loss functions for gradient boosting. We devise a spatial cross-validation scheme and show that in our setting it provides a better proxy for test set performance than naive cross-validation. The predictions are benchmarked against boosting approaches with different loss functions, and perform competitively in terms of the score criterion, finally placing second in the competition ranking.

An Auxiliary Approach to Prediction of Binary Outcome with Bayesian Network Model: Exploration with Data for Recurrence of Breast Cancer

Introduction: Logistic regression is the classical statistical model that is incorporated to predict a binary outcome variable. These models have theoretical assumptions of independence of predictor variables and linearity of association with the outcome in the logarithmic scale. Alternative models developed in the machine learning context like Naïve Bayes model with similar assumptions and Bayesian Network (BN) model can be used for binary prediction. Aim: To compare the predictive performance of logistic regression, Naïve Bayes and BN model in predicting the recurrence of Breast cancer. Materials and Methods: The dataset was procured from UCI Machine Learning repository on recurrence of breast cancer. The study was done on retrospective data from December 2021 to July 2022. The sample size was boosted with the bootstrapping with logistic regression model. The dataset was split into training (70%) and testing (30%) dataset for internal validation. The effect estimates of the potential prognostic variables were estimated using multiple logistic regression model. Naïve Bayes and BN model was also learnt from the training dataset. The indices of predictive accuracy were estimated for the models in both training and testing dataset. Results: Degree of malignancy and side of affected breast were found to be significant predictors of recurrence of breast cancer. BN model had the least misclassification rate and the best sensitivity in comparison to other models in spite of imbalance in outcome variable. Conclusion: BN model performed the best in comparison to logistic regression model when the assumptions of logistic regression model were violated and there is imbalance in proportion of outcome.