Abstract

Abstract
As machine learning technologies are increasingly adopted, the demand for responsible AI practices to ensure transparency and accountability grows. To better understand the decision-making processes of machine learning models, GASTeN was developed to generate realistic yet ambiguous synthetic data near a classifier's decision boundary. However, the results were inconsistent, with few images in the low-confidence region and noise. Therefore, we propose a new GASTeN version with a modified architecture and a novel loss function. This new loss function incorporates a multi-objective measure with a Gaussian loss centered on the classifier probability, targeting the decision boundary. Our study found that while the original GASTeN architecture yields the highest Frechet Inception Distance (FID) scores, the updated version achieves lower Average Confusion Distance (ACD) values and consistent performance across low-confidence regions. Both architectures produce realistic and ambiguous images, but the updated one is more reliable, with no instances of GAN mode collapse. Additionally, the introduction of the Gaussian loss enhanced this architecture by allowing for adjustable tolerance in image generation around the decision boundary.

Abstract
Recent state-of-the-art forecasting methods are trained on collections of time series. These methods, often referred to as global models, can capture common patterns in different time series to improve their generalization performance. However, they require large amounts of data that might not be available. Moreover, global models may fail to capture relevant patterns unique to a particular time series. In these cases, data augmentation can be useful to increase the sample size of time series datasets. The main contribution of this work is a novel method for generating univariate time series synthetic samples. Our approach stems from the insight that the observations concerning a particular time series of interest represent only a small fraction of all observations. In this context, we frame the problem of training a forecasting model as an imbalanced learning task. Oversampling strategies are popular approaches used to handle the imbalance problem in machine learning. We use these techniques to create synthetic time series observations and improve the accuracy of forecasting models. We carried out experiments using 7 different databases that contain a total of 5502 univariate time series. We found that the proposed solution outperforms both a global and a local model, thus providing a better trade-off between these two approaches.

Abstract
Hyper-parameter optimization in machine learning models is critical for achieving peak performance. Over the past few years, numerous researchers have worked on this optimization challenge. They primarily focused on batch learning tasks where data distributions remain relatively unchanged. However, addressing the properties of data streams poses a substantial challenge. With the rapid evolution of technology, the demand for sophisticated techniques to handle dynamic data streams is becoming increasingly urgent. This paper introduces a novel adaptation of the Fish School Search (FSS) Algorithm for online hyper-parameter optimization, the FSS-SPT. The FSS-SPT is a solution designed explicitly for the dynamic context of data streams. One fundamental property of the FSS-SPT is that it can change between exploration and exploitation modes to cope with the concept drift and converge to reasonable solutions. Our experiments on different datasets provide compelling evidence of the superior performance of our proposed methodology, the FSS-SPT. It outperformed existing algorithms in two machine learning tasks, demonstrating its potential for practical application.

Abstract
Hierarchical document classification is essential for structuring large-scale textual corpora in domains such as digital libraries and academic repositories. While recent advances in large language models (LLMs) have opened new possibilities for text classification, their applicability to hierarchical settings under real-world constraints remains underexplored. This study investigates both generative and discriminative transformer-based models, evaluating their effectiveness across multiple inference strategies: zero-shot baseline, local fine-tuning, and a global approach using category-specific models. Experiments on two real-world hierarchical datasets provide a comprehensive comparison of classification accuracy, F1-macro scores, and inference times. The results highlight that, although generative LLMs can deliver competitive (yet variable) performance at higher levels of the hierarchy, their high inference costs hinder their use in time-sensitive applications. In contrast, fine-tuned discriminative models—particularly BERT-based architectures—consistently offer a more favorable trade-off between performance and efficiency. © 2025 Elsevier B.V., All rights reserved.

Abstract
Traditional machine learning techniques assume that data is drawn from a stationary source. This assumption is challenged in contexts with data streams for presenting constant and potentially infinite sequences whose distribution is prone to change over time. Based on these settings, detecting changes (a.k.a. concept drifts) is necessary to keep learning models up-to-date. Although state-of-the-art detection methods were designed to monitor the loss of predictive models, such monitoring falls short in many real-world scenarios where the true labels are not readily available. Therefore, there is increasing attention to unsupervised concept drift detection methods as approached in this paper. In this work, we present an unsupervised and interpretable method based on Radial Basis Function Networks (RBFN) and Markov Chains (MC), referred to as RMIDDM (Radial Markov Interpretable Drift Detection Method). In our method, RBF performs, in the intermediate layer, an activation process that implicitly produces groups of observations collected over time. Simultaneously, MC models the transitions between groups to support the detection of concept drifts, which happens when the active group changes and its probability exceeds a given threshold. A set of experiments with synthetic datasets and comparisons with state-of-the-art algorithms demonstrated that the proposed method can detect drifts at runtime in an efficient, interpretable, and independent way of labels, presenting competitive results and behavior. Additionally, to show its applicability in a real-world scenario, we analyzed new COVID-19 cases, deaths, and vaccinations to identify new waves as concept drifts and generate Markov models that allow understanding of their interaction.