Jonathan D. Burnett, Kristin L. Jaeger, Sherri L. Johnson
et al.
Abstract Accurate mapping of headwater streams and their flow status has important implications for understanding and managing water resources and land uses. However, accurate information is rare, especially in rugged, forested terrain. We developed a streamflow permanence classification model for forested lands in western Oregon using the latest light detection and ranging‐derived hydrography published in the National Hydrography Dataset. Models were trained using 2,518 flow/no flow field observations collected in late summer 2019–2021 across headwaters of 129 sub‐watersheds. The final model, the Western Oregon WeT DRy model, used Random Forest and 13 environmental covariates for classifying every 5‐m stream sub‐reach across 426 sub‐watersheds. The most important covariates were annual precipitation and drainage area. Model output included probabilities of late summer surface flow presence and were subsequently categorized into three streamflow permanence classes—Wet, Dry, and Ambiguous. Ambiguous denoted model probabilities and associated prediction intervals that extended over the 50% classification threshold between wet and dry. Model accuracy was 0.83 for sub‐watersheds that contained training data and decreased to 0.67 for sub‐watersheds that did not have observations of late summer surface flow. The model identified where predictions extrapolated beyond the domain characterized by the training data. The combination of spatially continuous estimates of late summer streamflow status along with uncertainty and extrapolation estimates provide critical information for strategic project planning and designing additional field data collection.
Abdelmoula Seqqam, Meryam Touirsi, Saliha Najib
et al.
Growing water scarcity, driven by climate change, population growth, and expanding human activities, poses a critical challenge to arid and semi-arid regions worldwide. In Morocco, the Khouribga region illustrates this stress, where limited recharge, recurrent droughts, and intensive groundwater abstraction threaten long-term water security. To address these pressures, this study applied an integrated framework combining remote sensing, Geographic Information Systems (GIS) and the Analytic Hierarchy Process (AHP) to delineate groundwater potential zones (GWPZ). Eight hydrogeological parameters, namely rainfall, lithology, slope, lineament density, drainage density, land use and land cover, distance to rivers, and potential evapotranspiration, were weighted through AHP and integrated using the Weighted Linear Combination method. The resulting map shows low (24.97 %), moderate (49.94 %), high (24.81 %) and very high (0.25 %) potential areas. Validation with 72 wells and boreholes achieved 83.33 % concordance and R2 = 0.75, confirming model reliability. High-potential sectors in the north and northeast of Boujaad reflect favorable geological structures, fracture networks, and precipitation patterns. The results offer a practical basis for targeting drilling, designing artificial recharge systems, and protecting infiltration areas. Future work should incorporate higher-resolution hydrogeological data, extended climate series, and machine learning approaches to improve predictive performance and adaptability in other semi-arid contexts.
Soumyadeep Ray, Pronaya Bhattacharya, Ebrahim A. Mattar
et al.
This survey examines the convergence of Explainable Artificial Intelligence (XAI) and Quantum Computing (QC) toward precision medicine. We review developments from 2018 to 2025, summarizing quantum algorithms, quantum-machine-learning models and XAI techniques applied to drug discovery, disease diagnosis, patient monitoring and biomarker identification. We introduce a taxonomy of hybrid and quantum-explainable approaches, evaluate NISQ hardware and encoding constraints, and compare interpretability methods (SHAP, LIME, QSHAP, QLRP, TSBA). Two case studies (doxorubicin cardiotoxicity prediction and pre-symptomatic IBD flare forecasting) demonstrate hybrid variational-quantum pipelines wrapped with SHAP-based explanations. We identify practical barriers (noise, data encoding, regulation, privacy) and outline research directions to benchmark clinical quantum advantage and develop scalable, transparent QXAI frameworks. The survey aims to guide interdisciplinary efforts toward trustworthy, scalable quantum-enabled precision healthcare.
Research questionCan machine learning models accurately predict the risk of early miscarriage following single vitrified-warmed blastocyst transfer (SVBT)?DesignA dual-center retrospective analysis of 1,664 SVBT cycles, including 308 early miscarriage cases, was conducted across two reproductive centers. Multiple machine learning models, such as Logistic Regression, Random Forest, Gradient Boosting, and Voting Classifier, were developed. Metrics including Area Under the Curve(AUC), accuracy, precision, recall, F1 score, and specificity were used to evaluate model performance. Key predictors were identified through Mutual Information and Recursive Feature Elimination (RFE).ResultsMaternal age, paternal age, endometrial thickness, blastocyst quality, and ovarian stimulation parameters were identified as critical predictors. Compared to traditional statistical models such as logistic regression (AUC = 0.584), ensemble models demonstrated significantly improved predictive performance. The Voting Classifier achieved the highest AUC (0.836), accuracy (0.780), precision (0.914), and specificity (0.942), outperforming individual machine learning classifiers. The Gradient Boosting Classifier also exhibited strong performance (AUC 0.831, accuracy 0.777), confirming the effectiveness of ensemble learning in capturing complex predictors of early miscarriage risk.ConclusionEnsemble machine learning models, particularly the Voting Classifier and Gradient Boosting Classifier, significantly improve the prediction of early miscarriage following SVBT. These models provide accurate, individualized risk assessments, enhancing clinical decision-making and advancing personalized care in ART.
Diseases of the endocrine glands. Clinical endocrinology
Brain–computer interfaces (BCIs) enable people to communicate with others or devices, and improving BCI performance is essential for developing real-life applications. In this study, a steady-state visual evoked potential-based BCI (SSVEP-based BCI) with multi-domain features and multi-task learning is developed. To accurately represent the characteristics of an SSVEP signal, SSVEP signals in the time and frequency domains are selected as multi-domain features. Convolutional neural networks are separately used for time and frequency domain signals to extract the embedding features effectively. An element-wise addition operation and batch normalization are applied to fuse the time- and frequency-domain features. A sequence of convolutional neural networks is then adopted to find discriminative embedding features for classification. Finally, multi-task learning-based neural networks are used to detect the corresponding stimuli correctly. The experimental results showed that the proposed approach outperforms EEGNet, multi-task learning-based neural networks, canonical correlation analysis (CCA), and filter bank CCA (FBCCA). Additionally, the proposed approach is more suitable for developing real-time BCIs than a system where an input’s duration is 4 s. In the future, utilizing multi-task learning to learn the properties of the embedding features extracted from FBCCA can further improve the BCI system performance.
Arpit Guleria, Harshan Jagadeesh, Ranjitha Prasad
et al.
Abstract Class imbalance in training datasets can lead to bias and poor generalization in machine learning models. While pre-processing of training datasets is an efficient way to address both these issues in centralized learning environments, it is challenging to detect and address these issues in a distributed learning environment such as federated learning. In this paper, we propose FLICKER, a privacy preserving framework to address issues related to global class imbalance in federated learning. At the heart of our contribution lies the popular Cheon-Kim-Kim-Song (CKKS) homomorphic encryption scheme, which is used by the clients to privately share their data attributes, and subsequently balance their datasets before implementing the federated learning scheme. Extensive experimental results show that our proposed method improves the federated learning accuracy numbers by up to 8 $$\%$$ when used along with popular datasets and relevant baselines.
Information technology, Electronic computers. Computer science
Path planning enables autonomous agents such as robots, self-driving vehicles, and UAVs to navigate from a starting point to a target destination while avoiding obstacles and adhering to operational constraints. As autonomous technologies become more prevalent in real-world applications, the demand for robust, adaptive, and computationally efficient path planning algorithms has intensified. This paper presents a comprehensive review of path planning strategies, focusing on classical, metaheuristic, and AI-based approaches. It explores the challenges posed by dynamic environments, non-holonomic constraints, and varying levels of environmental knowledge. The review also examines the strengths and limitations of each algorithmic category, highlighting their suitability for diverse applications ranging from industrial automation to autonomous navigation. Furthermore, the paper discusses emerging trends, including the integration of machine learning and reinforcement learning techniques, and outlines future research directions aimed at enhancing the adaptability and performance of path planning systems in complex, unstructured environments.
Tensor Network (TN) Kernel Machines speed up model learning by representing parameters as low-rank TNs, reducing computation and memory use. However, most TN-based Kernel methods are deterministic and ignore parameter uncertainty. Further, they require manual tuning of model complexity hyperparameters like tensor rank and feature dimensions, often through trial-and-error or computationally costly methods like cross-validation. We propose Bayesian Tensor Network Kernel Machines, a fully probabilistic framework that uses sparsity-inducing hierarchical priors on TN factors to automatically infer model complexity. This enables automatic inference of tensor rank and feature dimensions, while also identifying the most relevant features for prediction, thereby enhancing model interpretability. All the model parameters and hyperparameters are treated as latent variables with corresponding priors. Given the Bayesian approach and latent variable dependencies, we apply a mean-field variational inference to approximate their posteriors. We show that applying a mean-field approximation to TN factors yields a Bayesian ALS algorithm with the same computational complexity as its deterministic counterpart, enabling uncertainty quantification at no extra computational cost. Experiments on synthetic and real-world datasets demonstrate the superior performance of our model in prediction accuracy, uncertainty quantification, interpretability, and scalability.
Systematic reviews (SRs) employ standardized methodological processes for synthesizing empirical evidence to answer specific research questions. These processes include rigorous screening phases to determine eligibility of articles against strict inclusion and exclusion criteria. Despite these processes, SRs are a significant undertaking, and this type of research often necessitates extensive human resource requirements, especially when the scope of the review is large. Given the substantial resources and time commitment required, we investigated a way in which the screening process might be accelerated while maintaining high fidelity and adherence to SR processes. More recently, researchers have turned to artificial intelligence-based (AI) software to expedite the screening process. This paper evaluated the agreement and usability of a novel machine learning program, Sciome SWIFT-ActiveScreener (ActiveScreener), in a large SR of mental health outcomes following treatment for PTSD. ActiveScreener exceeded the expected 95% agreement of the program with screeners to predict inclusion or exclusion of relevant articles at the title/abstract assessment phase of the review and was reported to be user friendly by both novice and seasoned screeners. ActiveScreener, when used appropriately, may be a useful tool when performing SR in a clinical context.
Morimasa Suzuna, Iwasaki Masako, Suehara Taikan
et al.
We have developed an energy calibration method using machine learning for the ILC electromagnetic (EM) calorimeter (ECAL), a sampling calorimeter consisting of Silicon-Tungsten layers. In this method, we use a deep neural network (DNN) for a regression to determine the energy of incident EM particles, improving the energy calibration resolution of the ECAL. The DNN architecture takes cluster hit data as low-level features of the cluster. In this paper, we report the status of our R&D and present results on energy calibration accuracy.
Brandon Warner, Edward Ratner, Kallin Carlous-Khan
et al.
This paper proposes a novel model-agnostic method for weighting the outputs of base classifiers in machine learning (ML) ensembles. Our approach uses class-based weight coefficients assigned to every output class in each learner in the ensemble. This is particularly useful when the base classifiers have highly variable performance across classes. Our method generates a dense set of coefficients for the models in our ensemble by considering the model performance on each class. We compare our novel method to the commonly used ensemble approaches like voting and weighted averages. In addition, we compare our approach to class-specific soft voting (CSSV), which was also designed to address variable performance but generates a sparse set of weights by solving a linear system. We choose to illustrate the power of this approach by applying it to an ensemble of extreme learning machines (ELMs), which are well suited for this approach due to their stochastic, highly varying performance across classes. We illustrate the superiority of our approach by comparing its performance to that of simple majority voting, weighted majority voting, and class-specific soft voting using ten popular open-source multiclass classification datasets.
In contrast with Mercer kernel-based approaches as used e.g., in Kernel Principal Component Analysis (KPCA), it was previously shown that Singular Value Decomposition (SVD) inherently relates to asymmetric kernels and Asymmetric Kernel Singular Value Decomposition (KSVD) has been proposed. However, the existing formulation to KSVD cannot work with infinite-dimensional feature mappings, the variational objective can be unbounded, and needs further numerical evaluation and exploration towards machine learning. In this work, i) we introduce a new asymmetric learning paradigm based on coupled covariance eigenproblem (CCE) through covariance operators, allowing infinite-dimensional feature maps. The solution to CCE is ultimately obtained from the SVD of the induced asymmetric kernel matrix, providing links to KSVD. ii) Starting from the integral equations corresponding to a pair of coupled adjoint eigenfunctions, we formalize the asymmetric Nyström method through a finite sample approximation to speed up training. iii) We provide the first empirical evaluations verifying the practical utility and benefits of KSVD and compare with methods resorting to symmetrization or linear SVD across multiple tasks.
Systems characterized by evolving interactions, prevalent in social, financial, and biological domains, are effectively modeled as continuous-time dynamic graphs (CTDGs). To manage the scale and complexity of these graph datasets, machine learning (ML) approaches have become essential. However, CTDGs pose challenges for ML because traditional static graph methods do not naturally account for event timings. Newer approaches, such as graph recurrent neural networks (GRNNs), are inherently time-aware and offer advantages over static methods for CTDGs. However, GRNNs face another issue: the short truncation of backpropagation-through-time (BPTT), whose impact has not been properly examined until now. In this work, we demonstrate that this truncation can limit the learning of dependencies beyond a single hop, resulting in reduced performance. Through experiments on a novel synthetic task and real-world datasets, we reveal a performance gap between full backpropagation-through-time (F-BPTT) and the truncated backpropagation-through-time (T-BPTT) commonly used to train GRNN models. We term this gap the "truncation gap" and argue that understanding and addressing it is essential as the importance of CTDGs grows, discussing potential future directions for research in this area.