arXiv subjects

Catarina Moreira

Publications and source records attributed to Catarina Moreira.

At least 19 recordsLinked to original sources

arXiv2026

Shattering the Shortcut: A Topology-Regularized Benchmark for Multi-hop Medical Reasoning in LLMs

While Large Language Models (LLMs) achieve expert-level performance on standard medical benchmarks through single-hop factual recall, they severely struggle with the complex, multi-hop diagnostic reasoning required in real-world clinical settings. A primary obstacle is "shortcut learning", where models exploit highly connected, generic hub nodes (e.g., "inflammation") in knowledge graphs to bypass authentic micro-pathological cascades. To address this, we introduce ShatterMed-QA, a bilingual benchmark of 10,558 multi-hop clinical questions designed to rigorously evaluate deep diagnostic reasoning. Our framework constructs a topology-regularized medical Knowledge Graph using a novel $k$-Shattering algorithm, which physically prunes generic hubs to explicitly sever logical shortcuts. We synthesize the evaluation vignettes by applying implicit bridge entity masking and topology-driven hard negative sampling, forcing models to navigate biologically plausible distractors without relying on superficial elimination. Comprehensive evaluations of 21 LLMs reveal massive performance degradation on our multi-hop tasks, particularly among domain-specific models. Crucially, restoring the masked evidence via Retrieval-Augmented Generation (RAG) triggers near-universal performance recovery, validating ShatterMed-QA's structural fidelity and proving its efficacy in diagnosing the fundamental reasoning deficits of current medical AI. Explore the dataset, interactive examples, and full leaderboards at our project website: https://shattermed-qa-web.vercel.app/

cs.CL↗

arXiv2025

Crafting Imperceptible On-Manifold Adversarial Attacks for Tabular Data

Adversarial attacks on tabular data present unique challenges due to the heterogeneous nature of mixed categorical and numerical features. Unlike images where pixel perturbations maintain visual similarity, tabular data lacks intuitive similarity metrics, making it difficult to define imperceptible modifications. Additionally, traditional gradient-based methods prioritise $\ell_p$-norm constraints, often producing adversarial examples that deviate from the original data distributions. To address this, we propose a latent-space perturbation framework using a mixed-input Variational Autoencoder (VAE) to generate statistically consistent adversarial examples. The proposed VAE integrates categorical embeddings and numerical features into a unified latent manifold, enabling perturbations that preserve statistical consistency. We introduce In-Distribution Success Rate (IDSR) to jointly evaluate attack effectiveness and distributional alignment. Evaluation across six publicly available datasets and three model architectures demonstrates that our method achieves substantially lower outlier rates and more consistent performance compared to traditional input-space attacks and other VAE-based methods adapted from image domain approaches, achieving substantially lower outlier rates and higher IDSR across six datasets and three model architectures. Our comprehensive analyses of hyperparameter sensitivity, sparsity control, and generative architecture demonstrate that the effectiveness of VAE-based attacks depends strongly on reconstruction quality and the availability of sufficient training data. When these conditions are met, the proposed framework achieves superior practical utility and stability compared with input-space methods. This work underscores the importance of maintaining on-manifold perturbations for generating realistic and robust adversarial examples in tabular domains.

cs.LG↗

arXiv2025

HeadZoom: Hands-Free Zooming and Panning for 2D Image Navigation Using Head Motion

We introduce \textit{HeadZoom}, a hands-free interaction technique for navigating two-dimensional visual content using head movements. HeadZoom enables fluid zooming and panning using only real-time head tracking. It supports natural control in applications such as map exploration, radiograph inspection, and image browsing, where physical interaction is limited. We evaluated HeadZoom in a within-subjects study comparing three interaction techniques-Static, Tilt Zoom, and Parallel Zoom-across spatial, error, and subjective metrics. Parallel Zoom significantly reduced total head movement compared to Static and Tilt modes. Users reported significantly lower perceived exertion for Parallel Zoom, confirming its suitability for prolonged or precision-based tasks. By minimizing movement demands while maintaining task effectiveness, HeadZoom advances the design of head-based 2D interaction in VR and creates new opportunities for accessible hands-free systems for image exploration.

cs.HC↗

arXiv2025

Explainable AI Systems Must Be Contestable: Here's How to Make It Happen

As AI regulations around the world intensify their focus on system safety, contestability has become a mandatory, yet ill-defined, safeguard. In XAI, "contestability" remains an empty promise: no formal definition exists, no algorithm guarantees it, and practitioners lack concrete guidance to satisfy regulatory requirements. Grounded in a systematic literature review, this paper presents the first rigorous formal definition of contestability in explainable AI, directly aligned with stakeholder requirements and regulatory mandates. We introduce a modular framework of by-design and post-hoc mechanisms spanning human-centered interfaces, technical architectures, legal processes, and organizational workflows. To operationalize our framework, we propose the Contestability Assessment Scale, a composite metric built on more than twenty quantitative criteria. Through multiple case studies across diverse application domains, we reveal where state-of-the-art systems fall short and show how our framework drives targeted improvements. By converting contestability from regulatory theory into a practical framework, our work equips practitioners with the tools to embed genuine recourse and accountability into AI systems.

cs.CY↗

arXiv2025

TabAttackBench: A Benchmark for Adversarial Attacks on Tabular Data

Adversarial attacks pose a significant threat to machine learning models by inducing incorrect predictions through imperceptible perturbations to input data. While these attacks are well studied in unstructured domains such as images, their behaviour on tabular data remains underexplored due to mixed feature types and complex inter-feature dependencies. This study introduces a comprehensive benchmark that evaluates adversarial attacks on tabular datasets with respect to both effectiveness and imperceptibility. We assess five white-box attack algorithms (FGSM, BIM, PGD, DeepFool, and C\&W) across four representative models (LR, MLP, TabTransformer and FT-Transformer) using eleven datasets spanning finance, energy, and healthcare domains. The benchmark employs four quantitative imperceptibility metrics (proximity, sparsity, deviation, and sensitivity) to characterise perturbation realism. The analysis quantifies the trade-off between these two aspects and reveals consistent differences between attack types, with $\ell_\infty$-based attacks achieving higher success but lower subtlety, and $\ell_2$-based attacks offering more realistic perturbations. The benchmark findings offer actionable insights for designing more imperceptible adversarial attacks, advancing the understanding of adversarial vulnerability in tabular machine learning.

cs.LG↗

arXiv2025

Curation and Analysis of MIMICEL -- An Event Log for MIMIC-IV Emergency Department

The global issue of overcrowding in emergency departments (ED) necessitates the analysis of patient flow through ED to enhance efficiency and alleviate overcrowding. However, traditional analytical methods are time-consuming and costly. The healthcare industry is embracing process mining tools to analyse healthcare processes and patient flows. Process mining aims to discover, monitor, and enhance processes by obtaining knowledge from event log data. However, the availability of event logs is a prerequisite for applying process mining techniques. Hence, this paper aims to generate an event log for analysing processes in ED. In this study, we extract an event log from the MIMIC-IV-ED dataset and name it MIMICEL. MIMICEL captures the process of patient journey in ED, allowing for analysis of patient flows and improving ED efficiency. We present analyses conducted using MIMICEL to demonstrate the utility of the dataset. The curation of MIMICEL facilitates extensive use of MIMIC-IV-ED data for ED analysis using process mining techniques, while also providing the process mining research communities with a valuable dataset for study.

cs.DB↗

arXiv2025

SPARC: Shared Perspective with Avatar Distortion for Remote Collaboration in VR

Telepresence VR systems allow for face-to-face communication, promoting the feeling of presence and understanding of nonverbal cues. However, when discussing virtual 3D objects, limitations to presence and communication cause deictic gestures to lose meaning due to disparities in orientation. Current approaches use shared perspective, and avatar overlap to restore these references, which cause occlusions and discomfort that worsen when multiple users participate. We introduce a new approach to shared perspective in multi-user collaboration where the avatars are not co-located. Each person sees the others' avatars at their positions around the workspace while having a first-person view of the workspace. Whenever a user manipulates an object, others will see his/her arms stretching to reach that object in their perspective. SPARC combines a shared orientation and supports nonverbal communication, minimizing occlusions. We conducted a user study (n=18) to understand how the novel approach impacts task performance and workspace awareness. We found evidence that SPARC is more efficient and less mentally demanding than life-like settings.

cs.HC↗

arXiv2025

Gaze-Hand Steering for Travel and Multitasking in Virtual Environments

As head-mounted displays (HMDs) with eye-tracking become increasingly accessible, the need for effective gaze-based interfaces in virtual reality (VR) grows. Traditional gaze- or hand-based navigation often limits user precision or impairs free viewing, making multitasking difficult. We present a gaze-hand steering technique that combines eye-tracking with hand-pointing: users steer only when gaze aligns with a hand-defined target, reducing unintended actions and enabling free look. Speed is controlled via either a joystick or a waist-level speed circle. We evaluated our method in a user study (N=20) across multitasking and single-task scenarios, comparing it to a similar technique. Results show that gaze-hand steering maintains performance and enhances user comfort and spatial awareness during multitasking. Our findings support the use of gaze-hand steering in gaze-dominant VR applications requiring precision and simultaneous interaction. Our method significantly improves VR navigation in gaze-dominant, multitasking-intensive applications, supporting immersion and efficient control.

cs.HC↗

arXiv2025

DALL-M: Context-Aware Clinical Data Augmentation with LLMs

X-ray images are vital in medical diagnostics, but their effectiveness is limited without clinical context. Radiologists often find chest X-rays insufficient for diagnosing underlying diseases, necessitating the integration of structured clinical features with radiology reports. To address this, we introduce DALL-M, a novel framework that enhances clinical datasets by generating contextual synthetic data. DALL-M augments structured patient data, including vital signs (e.g., heart rate, oxygen saturation), radiology findings (e.g., lesion presence), and demographic factors. It integrates this tabular data with contextual knowledge extracted from radiology reports and domain-specific resources (e.g., Radiopaedia, Wikipedia), ensuring clinical consistency and reliability. DALL-M follows a three-phase process: (i) clinical context storage, (ii) expert query generation, and (iii) context-aware feature augmentation. Using large language models (LLMs), it generates both contextual synthetic values for existing clinical features and entirely new, clinically relevant features. Applied to 799 cases from the MIMIC-IV dataset, DALL-M expanded the original 9 clinical features to 91. Empirical validation with machine learning models (including Decision Trees, Random Forests, XGBoost, and TabNET) demonstrated a 16.5% improvement in F1 score and a 25% increase in Precision and Recall. DALL-M bridges an important gap in clinical data augmentation by preserving data integrity while enhancing predictive modeling in healthcare. Our results show that integrating LLM-generated synthetic features significantly improves model performance, making DALL-M a scalable and practical approach for AI-driven medical diagnostics.

cs.AI↗

arXiv2025

Weakly Supervised Pixel-Level Annotation with Visual Interpretability

Medical image annotation is essential for diagnosing diseases, yet manual annotation is time-consuming, costly, and prone to variability among experts. To address these challenges, we propose an automated explainable annotation system that integrates ensemble learning, visual explainability, and uncertainty quantification. Our approach combines three pre-trained deep learning models - ResNet50, EfficientNet, and DenseNet - enhanced with XGrad-CAM for visual explanations and Monte Carlo Dropout for uncertainty quantification. This ensemble mimics the consensus of multiple radiologists by intersecting saliency maps from models that agree on the diagnosis while uncertain predictions are flagged for human review. We evaluated our system using the TBX11K medical imaging dataset and a Fire segmentation dataset, demonstrating its robustness across different domains. Experimental results show that our method outperforms baseline models, achieving 93.04% accuracy on TBX11K and 96.4% accuracy on the Fire dataset. Moreover, our model produces precise pixel-level annotations despite being trained with only image-level labels, achieving Intersection over Union IoU scores of 36.07% and 64.7%, respectively. By enhancing the accuracy and interpretability of image annotations, our approach offers a reliable and transparent solution for medical diagnostics and other image analysis tasks.

cs.CV↗

arXiv2025

TagGAN: A Generative Model for Data Tagging

Precise identification and localization of disease-specific features at the pixel-level are particularly important for early diagnosis, disease progression monitoring, and effective treatment in medical image analysis. However, conventional diagnostic AI systems lack decision transparency and cannot operate well in environments where there is a lack of pixel-level annotations. In this study, we propose a novel Generative Adversarial Networks (GANs)-based framework, TagGAN, which is tailored for weakly-supervised fine-grained disease map generation from purely image-level labeled data. TagGAN generates a pixel-level disease map during domain translation from an abnormal image to a normal representation. Later, this map is subtracted from the input abnormal image to convert it into its normal counterpart while preserving all the critical anatomical details. Our method is first to generate fine-grained disease maps to visualize disease lesions in a weekly supervised setting without requiring pixel-level annotations. This development enhances the interpretability of diagnostic AI by providing precise visualizations of disease-specific regions. It also introduces automated binary mask generation to assist radiologists. Empirical evaluations carried out on the benchmark datasets, CheXpert, TBX11K, and COVID-19, demonstrate the capability of TagGAN to outperform current top models in accurately identifying disease-specific pixels. This outcome highlights the capability of the proposed model to tag medical images, significantly reducing the workload for radiologists by eliminating the need for binary masks during training.

eess.IV↗

arXiv2024

PROMPTHEUS: A Human-Centered Pipeline to Streamline SLRs with LLMs

The growing volume of academic publications poses significant challenges for researchers conducting timely and accurate Systematic Literature Reviews, particularly in fast-evolving fields like artificial intelligence. This growth of academic literature also makes it increasingly difficult for lay people to access scientific knowledge effectively, meaning academic literature is often misrepresented in the popular press and, more broadly, in society. Traditional SLR methods are labor-intensive and error-prone, and they struggle to keep up with the rapid pace of new research. To address these issues, we developed \textit{PROMPTHEUS}: an AI-driven pipeline solution that automates the SLR process using Large Language Models. We aimed to enhance efficiency by reducing the manual workload while maintaining the precision and coherence required for comprehensive literature synthesis. PROMPTHEUS automates key stages of the SLR process, including systematic search, data extraction, topic modeling using BERTopic, and summarization with transformer models. Evaluations conducted across five research domains demonstrate that PROMPTHEUS reduces review time, achieves high precision, and provides coherent topic organization, offering a scalable and effective solution for conducting literature reviews in an increasingly crowded research landscape. In addition, such tools may reduce the increasing mistrust in science by making summarization more accessible to laypeople. The code for this project can be found on the GitHub repository at https://github.com/joaopftorres/PROMPTHEUS.git

cs.AI↗

arXiv2024

Investigating Imperceptibility of Adversarial Attacks on Tabular Data: An Empirical Analysis

Adversarial attacks are a potential threat to machine learning models by causing incorrect predictions through imperceptible perturbations to the input data. While these attacks have been extensively studied in unstructured data like images, applying them to tabular data, poses new challenges. These challenges arise from the inherent heterogeneity and complex feature interdependencies in tabular data, which differ from the image data. To account for this distinction, it is necessary to establish tailored imperceptibility criteria specific to tabular data. However, there is currently a lack of standardised metrics for assessing the imperceptibility of adversarial attacks on tabular data. To address this gap, we propose a set of key properties and corresponding metrics designed to comprehensively characterise imperceptible adversarial attacks on tabular data. These are: proximity to the original input, sparsity of altered features, deviation from the original data distribution, sensitivity in perturbing features with narrow distribution, immutability of certain features that should remain unchanged, feasibility of specific feature values that should not go beyond valid practical ranges, and feature interdependencies capturing complex relationships between data attributes. We evaluate the imperceptibility of five adversarial attacks, including both bounded attacks and unbounded attacks, on tabular data using the proposed imperceptibility metrics. The results reveal a trade-off between the imperceptibility and effectiveness of these attacks. The study also identifies limitations in current attack algorithms, offering insights that can guide future research in the area. The findings gained from this empirical analysis provide valuable direction for enhancing the design of adversarial attack algorithms, thereby advancing adversarial machine learning on tabular data.

cs.LG↗

arXiv2024

Developing Guidelines for Functionally-Grounded Evaluation of Explainable Artificial Intelligence using Tabular Data

Explainable Artificial Intelligence (XAI) techniques are used to provide transparency to complex, opaque predictive models. However, these techniques are often designed for image and text data, and it is unclear how fit-for-purpose they are when applied to tabular data. As XAI techniques are rarely evaluated in settings with tabular data, the applicability of existing evaluation criteria and methods are also unclear and needs (re-)examination. For example, some works suggest that evaluation methods may unduly influence the evaluation results when using tabular data. This lack of clarity on evaluation procedures can lead to reduced transparency and ineffective use of XAI techniques in real world settings. In this study, we examine literature on XAI evaluation to derive guidelines on functionally-grounded assessment of local, post hoc XAI techniques. We identify 20 evaluation criteria and associated evaluation methods, and derive guidelines on when and how each criterion should be evaluated. We also identify key research gaps to be addressed by future work. Our study contributes to the body of knowledge on XAI evaluation through in-depth examination of functionally-grounded XAI evaluation protocols, and has laid the groundwork for future research on XAI evaluation.

cs.CY↗

arXiv2024

Benchmarking Instance-Centric Counterfactual Algorithms for XAI: From White Box to Black Box

This study investigates the impact of machine learning models on the generation of counterfactual explanations by conducting a benchmark evaluation over three different types of models: a decision tree (fully transparent, interpretable, white-box model), a random forest (semi-interpretable, grey-box model), and a neural network (fully opaque, black-box model). We tested the counterfactual generation process using four algorithms (DiCE, WatcherCF, prototype, and GrowingSpheresCF) in the literature in 25 different datasets. Our findings indicate that: (1) Different machine learning models have little impact on the generation of counterfactual explanations; (2) Counterfactual algorithms based uniquely on proximity loss functions are not actionable and will not provide meaningful explanations; (3) One cannot have meaningful evaluation results without guaranteeing plausibility in the counterfactual generation. Algorithms that do not consider plausibility in their internal mechanisms will lead to biased and unreliable conclusions if evaluated with the current state-of-the-art metrics; (4) A counterfactual inspection analysis is strongly recommended to ensure a robust examination of counterfactual explanations and the potential identification of biases.

cs.LG↗

arXiv2024

SelfReDepth: Self-Supervised Real-Time Depth Restoration for Consumer-Grade Sensors

Depth maps produced by consumer-grade sensors suffer from inaccurate measurements and missing data from either system or scene-specific sources. Data-driven denoising algorithms can mitigate such problems. However, they require vast amounts of ground truth depth data. Recent research has tackled this limitation using self-supervised learning techniques, but it requires multiple RGB-D sensors. Moreover, most existing approaches focus on denoising single isolated depth maps or specific subjects of interest, highlighting a need for methods to effectively denoise depth maps in real-time dynamic environments. This paper extends state-of-the-art approaches for depth-denoising commodity depth devices, proposing SelfReDepth, a self-supervised deep learning technique for depth restoration, via denoising and hole-filling by inpainting full-depth maps captured with RGB-D sensors. The algorithm targets depth data in video streams, utilizing multiple sequential depth frames coupled with color data to achieve high-quality depth videos with temporal coherence. Finally, SelfReDepth is designed to be compatible with various RGB-D sensors and usable in real-time scenarios as a pre-processing step before applying other depth-dependent algorithms. Our results demonstrate our approach's real-time performance on real-world datasets. They show that it outperforms state-of-the-art denoising and restoration performance at over 30fps on Commercial Depth Cameras, with potential benefits for augmented and mixed-reality applications.

cs.CV↗

arXiv2024

Bayesian Adaptive Trials for Social Policy

This paper proposes Bayesian Adaptive Trials (BAT) as both an efficient method to conduct trials and a unifying framework for evaluation social policy interventions, addressing limitations inherent in traditional methods such as Randomized Controlled Trials (RCT). Recognizing the crucial need for evidence-based approaches in public policy, the proposal aims to lower barriers to the adoption of evidence-based methods and align evaluation processes more closely with the dynamic nature of policy cycles. BATs, grounded in decision theory, offer a dynamic, ``learning as we go'' approach, enabling the integration of diverse information types and facilitating a continuous, iterative process of policy evaluation. BATs' adaptive nature is particularly advantageous in policy settings, allowing for more timely and context-sensitive decisions. Moreover, BATs' ability to value potential future information sources positions it as an optimal strategy for sequential data acquisition during policy implementation. While acknowledging the assumptions and models intrinsic to BATs, such as prior distributions and likelihood functions, the paper argues that these are advantageous for decision-makers in social policy, effectively merging the best features of various methodologies.

stat.AP↗

arXiv2024

Generating Feasible and Plausible Counterfactual Explanations for Outcome Prediction of Business Processes

In recent years, various machine and deep learning architectures have been successfully introduced to the field of predictive process analytics. Nevertheless, the inherent opacity of these algorithms poses a significant challenge for human decision-makers, hindering their ability to understand the reasoning behind the predictions. This growing concern has sparked the introduction of counterfactual explanations, designed as human-understandable what if scenarios, to provide clearer insights into the decision-making process behind undesirable predictions. The generation of counterfactual explanations, however, encounters specific challenges when dealing with the sequential nature of the (business) process cases typically used in predictive process analytics. Our paper tackles this challenge by introducing a data-driven approach, REVISEDplus, to generate more feasible and plausible counterfactual explanations. First, we restrict the counterfactual algorithm to generate counterfactuals that lie within a high-density region of the process data, ensuring that the proposed counterfactuals are realistic and feasible within the observed process data distribution. Additionally, we ensure plausibility by learning sequential patterns between the activities in the process cases, utilising Declare language templates. Finally, we evaluate the properties that define the validity of counterfactuals.

cs.AI↗