Feature-importance methods show promise in transforming machine learning models from predictive engines into tools for scientific discovery. However, due to data sampling and algorithmic stochasticity, expressive models can be unstable, leading to inaccurate variable importance estimates and undermining their utility in critical biomedical applications. Although ensembling offers a solution, deciding whether to explain a single ensemble model or aggregate individual model explanations is difficult due to the nonlinearity of importance measures and remains largely understudied. Our theoretical analysis, developed under assumptions accommodating complex state-of-the-art ML models, reveals that this choice is primarily driven by the model's excess risk. In contrast to prior literature, we show that ensembling at the model level provides more accurate variable-importance estimates, particularly for expressive models, by reducing this leading error term. We validate these findings on classical benchmarks and a large-scale proteomic study from the UK Biobank.

Abstract--The widespread adoption of complex machine learning models in high-stakes domains has brought the "black-box" problem to the forefront of responsible AI research. This paper aims at addressing this issue by improving the Explainable Boosting Machine (EBM), a state-of-the-art glassbox model that delivers both high accuracy and complete transparency. The paper outlines three distinct enhancement methodologies: targeted hyperparameter optimization with Bayesian methods, the implementation of a custom multi-objective function for fairness for hyperparameter optimization, and a novel self-supervised pre-training pipeline for cold-start scenarios. All three methodologies are evaluated across standard benchmark datasets, including the Adult Income, Credit Card Fraud Detection, and UCI Heart Disease datasets. The analysis indicates that while the tuning process yielded marginal improvements in the primary ROC AUC metric, it led to a subtle but important shift in the model's decision-making behavior, demonstrating the value of a multi-faceted evaluation beyond a single performance score. This work is positioned as a critical step toward developing machine learning systems that are not only accurate but also robust, equitable, and transparent, meeting the growing demands of regulatory and ethical compliance. A. The Black-Box Problem in High-Stakes Domains The remarkable surge in the performance of machine learning models has led to their pervasive adoption across a multitude of domains, from retail and finance to medicine and judicial systems. Complex, high-performing models, such as deep neural networks and ensemble methods like Random Forest and XGBoost, have become the de facto standard for many tasks.

Machine learning (ML) algorithms are heavily based on the availability of training data, which, depending on the domain, often includes sensitive information about data providers. This raises critical privacy concerns. Anonymization techniques have emerged as a practical solution to address these issues by generalizing features or suppressing data to make it more difficult to accurately identify individuals. Although recent studies have shown that privacy-enhancing technologies can influence ML predictions across different subgroups, thus affecting fair decision-making, the specific effects of anonymization techniques, such as $k$-anonymity, $\ell$-diversity, and $t$-closeness, on ML fairness remain largely unexplored. In this work, we systematically audit the impact of anonymization techniques on ML fairness, evaluating both individual and group fairness. Our quantitative study reveals that anonymization can degrade group fairness metrics by up to fourfold. Conversely, similarity-based individual fairness metrics tend to improve under stronger anonymization, largely as a result of increased input homogeneity. By analyzing varying levels of anonymization across diverse privacy settings and data distributions, this study provides critical insights into the trade-offs between privacy, fairness, and utility, offering actionable guidelines for responsible AI development. Our code is publicly available at: https://github.com/hharcolezi/anonymity-impact-fairness.

We examine frugal federated learning approaches to violence detection by comparing two complementary strategies: (i) zero-shot and federated fine-tuning of vision-language models (VLMs), and (ii) personalized training of a compact 3D convolutional neural network (CNN3D). Using LLaVA-7B and a 65.8M parameter CNN3D as representative cases, we evaluate accuracy, calibration, and energy usage under realistic non-IID settings. Both approaches exceed 90% accuracy. CNN3D slightly outperforms Low-Rank Adaptation(LoRA)-tuned VLMs in ROC AUC and log loss, while using less energy. VLMs remain favorable for contextual reasoning and multimodal inference. We quantify energy and CO$_2$ emissions across training and inference, and analyze sustainability trade-offs for deployment. To our knowledge, this is the first comparative study of LoRA-tuned vision-language models and personalized CNNs for federated violence detection, with an emphasis on energy efficiency and environmental metrics. These findings support a hybrid model: lightweight CNNs for routine classification, with selective VLM activation for complex or descriptive scenarios. The resulting framework offers a reproducible baseline for responsible, resource-aware AI in video surveillance, with extensions toward real-time, multimodal, and lifecycle-aware systems.

Data is vital in enabling machine learning models to advance research and practical applications in finance, where accurate and robust models are essential for investment and trading decision-making. However, real-world data is limited despite its quantity, quality, and variety. The data shortage of various financial assets directly hinders the performance of machine learning models designed to trade and invest in these assets. Generative methods can mitigate this shortage. In this paper, we introduce a set of novel techniques for time series data generation (we name them Fiaingen) and assess their performance across three criteria: (a) overlap of real-world and synthetic data on a reduced dimensionality space, (b) performance on downstream machine learning tasks, and (c) runtime performance. Our experiments demonstrate that the methods achieve state-of-the-art performance across the three criteria listed above. Synthetic data generated with Fiaingen methods more closely mirrors the original time series data while keeping data generation time close to seconds - ensuring the scalability of the proposed approach. Furthermore, models trained on it achieve performance close to those trained with real-world data.

We tackle atypical versus normal mitosis classification in the MIDOG 2025 challenge using EfficientViT-L2, a hybrid CNN--ViT architecture optimized for accuracy and efficiency. A unified dataset of 13,938 nuclei from seven cancer types (MIDOG++ and AMi-Br) was used, with atypical mitoses comprising ~15. To assess domain generalization, we applied leave-one-cancer-type-out cross-validation with 5-fold ensembles, using stain-deconvolution for image augmentation. For challenge submissions, we trained an ensemble with the same 5-fold split but on all cancer types. In the preliminary evaluation phase, this model achieved balanced accuracy of 0.859, ROC AUC of 0.942, and raw accuracy of 0.85, demonstrating competitive and well-balanced performance across metrics.

Machine learning (ML) is increasingly employed in real-world applications like medicine or economics, thus, potentially affecting large populations. However, ML models often do not perform homogeneously, leading to underperformance or, conversely, unusually high performance in certain subgroups (e.g., sex=female AND marital_status=married). Identifying such subgroups can support practical decisions on which subpopulation a model is safe to deploy or where more training data is required. However, an efficient and coherent framework for effective search is missing. Consequently, we introduce SubROC, an open-source, easy-to-use framework based on Exceptional Model Mining for reliably and efficiently finding strengths and weaknesses of classification models in the form of interpretable population subgroups. SubROC incorporates common evaluation measures (ROC and PR AUC), efficient search space pruning for fast exhaustive subgroup search, control for class imbalance, adjustment for redundant patterns, and significance testing. We illustrate the practical benefits of SubROC in case studies as well as in comparative analyses across multiple datasets.

The non-invasive assessment of increasingly incidentally discovered renal masses is a critical challenge in urologic oncology, where diagnostic uncertainty frequently leads to the overtreatment of benign or indolent tumors. In this study, we developed and validated RenalCLIP using a dataset of 27,866 CT scans from 8,809 patients across nine Chinese medical centers and the public TCIA cohort, a visual-language foundation model for characterization, diagnosis and prognosis of renal mass. The model was developed via a two-stage pre-training strategy that first enhances the image and text encoders with domain-specific knowledge before aligning them through a contrastive learning objective, to create robust representations for superior generalization and diagnostic precision. RenalCLIP achieved better performance and superior generalizability across 10 core tasks spanning the full clinical workflow of kidney cancer, including anatomical assessment, diagnostic classification, and survival prediction, compared with other state-of-the-art general-purpose CT foundation models. Especially, for complicated task like recurrence-free survival prediction in the TCIA cohort, RenalCLIP achieved a C-index of 0.726, representing a substantial improvement of approximately 20% over the leading baselines. Furthermore, RenalCLIP's pre-training imparted remarkable data efficiency; in the diagnostic classification task, it only needs 20% training data to achieve the peak performance of all baseline models even after they were fully fine-tuned on 100% of the data. Additionally, it achieved superior performance in report generation, image-text retrieval and zero-shot diagnosis tasks. Our findings establish that RenalCLIP provides a robust tool with the potential to enhance diagnostic accuracy, refine prognostic stratification, and personalize the management of patients with kidney cancer.

Understanding and classifying human cognitive brain states based on neuroimaging data remains one of the foremost and most challenging problems in neuroscience, owing to the high dimensionality and intrinsic noise of the signals. In this work, we propose an ensemble-based graph representation method of functional magnetic resonance imaging (fMRI) data for the task of binary brain-state classification. Our method builds the graph by leveraging multiple base machine-learning models: each edge weight reflects the difference in posterior probabilities between two cognitive states, yielding values in the range [-1, 1] that encode confidence in a given state. We applied this approach to seven cognitive tasks from the Human Connectome Project (HCP 1200 Subject Release), including working memory, gambling, motor activity, language, social cognition, relational processing, and emotion processing. Using only the mean incident edge weights of the graphs as features, a simple logistic-regression classifier achieved average accuracies from 97.07% to 99.74%. We also compared our ensemble graphs with classical correlation-based graphs in a classification task with a graph neural network (GNN). In all experiments, the highest classification accuracy was obtained with ensemble graphs. These results demonstrate that ensemble graphs convey richer topological information and enhance brain-state discrimination. Our approach preserves edge-level interpretability of the fMRI graph representation, is adaptable to multiclass and regression tasks, and can be extended to other neuroimaging modalities and pathological-state classification.

--The reliability of machine learning (ML) software systems is heavily influenced by changes in data over time. For that reason, ML systems require regular maintenance, typically based on model retraining. However, retraining requires significant computational demand, which makes it energy-intensive and raises concerns about its environmental impact. T o understand which retraining techniques should be considered when designing sustainable ML applications, in this work, we study the energy consumption of common retraining techniques. Since the accuracy of ML systems is also essential, we compare retraining techniques in terms of both energy efficiency and accuracy. We showcase that retraining with only the most recent data, compared to all available data, reduces energy consumption by up to 25%, being a sustainable alternative to the status quo. Furthermore, our findings show that retraining a model only when there is evidence that updates are necessary, rather than on a fixed schedule, can reduce energy consumption by up to 40%, provided a reliable data change detector is in place. Our findings pave the way for better recommendations for ML practitioners, guiding them toward more energy-efficient retraining techniques when designing sustainable ML software systems. The increasing adoption of Machine Learning (ML) and Artificial Intelligence (AI) within organizations has resulted in the development of more ML/AI software systems [1]. Although ML/AI brings plenty of business value, it is known that the accuracy of ML applications decreases over time [2]. Thus, ML developers must monitor and maintain their ML systems in production. One reason for this phenomenon is the fact that ML applications are highly dependent on the data on which they have been trained. Real-world data usually changes over time [3] - a phenomenon often referred to as concept drift [4] - which can significantly impact the normal operation of ML systems [5]. Therefore, appropriate maintenance techniques are required for the design of ML software systems. One common approach to maintaining these systems is to periodically update these applications by retraining the underlying ML models with the latest version of the data [6], [7]. On another note, the process of training machine learning models has raised substantial concerns about the carbon footprint of ML applications [8], [9].