arXiv subjects

Fabricio Murai

Publications and source records attributed to Fabricio Murai.

At least 19 recordsLinked to original sources

arXiv2026

CodeEvolve: an open source evolutionary coding agent for algorithmic discovery and optimization

We introduce CodeEvolve, an open-source framework that couples large language models with island-based evolutionary search for end-to-end algorithmic discovery. CodeEvolve integrates inspiration-based crossover, meta-prompting, and depth-based refinement on top of a CVT-MAP-Elites archive and a weighted LLM ensemble to generate optimized solutions for complex problems. On the AlphaEvolve benchmark suite, CodeEvolve matches or surpasses the reported AlphaEvolve results on 5 of 9 problems and, under matched conditions, outperforms the open-source frameworks OpenEvolve and ShinkaEvolve on 6 of 9. With the open-weight Qwen3-Coder-30B backbone, it surpasses the reported AlphaEvolve score on both CirclePackingSquare instances at roughly an order of magnitude lower cost than a frontier closed-source ensemble, and remains competitive with EoH on heuristic-design tasks without retuning. Ablations show that the interaction between CodeEvolve's components, rather than any single operator, drives these results. We release the framework, experimental data, and practical hyperparameter guidelines at https://github.com/inter-co/science-codeevolve.

cs.AI↗

arXiv2026

The Brazilian Vaccination Debate on YouTube: Topics, Perspectives, and Engagement Dynamics

Vaccination debates are central to online public health communication, as COVID-19 intensified disputes over scientific authority, institutional trust, and political identity. Yet studies often isolate semantic structure, stance, misinformation, and engagement, leaving their interplay over time poorly understood. We conduct a multilevel computational text analysis based on language models applied to 1.27 million Brazilian YouTube comments from 2018 to 2024, using what is, to our knowledge, the largest dataset of Brazilian vaccine discourse on the Web. We contrast producer framing in titles with audience discourse in comments, integrating Topic-derived themes with engagement metadata, conversational timing, stance-derived vaccine positions, and pre-pandemic, pandemic, and post-pandemic periods. Results show that COVID-19 dominates biomedical and informational themes in titles, whereas comments span personal health reports, vaccine effects, information credibility, conspiracy narratives, and political disputes. Health-related macro-topics dominate in scale and persistence, while conspiratorial and political themes are associated with faster interactions and a greater concentration of vaccine-opposing engagement. Post-pandemic activity remains centered on health experiences, vaccine effects, and information credibility, indicating no return to the pre-pandemic thematic configuration. By integrating semantic, interactional, stance, and temporal dimensions, this study shows how audiences reframe producer-framed health content and how vaccine controversies persist beyond the acute pandemic period.

cs.SI↗

arXiv2026

PubMed Reasoner: Dynamic Reasoning-based Retrieval for Evidence-Grounded Biomedical Question Answering

Trustworthy biomedical question answering (QA) systems must not only provide accurate answers but also justify them with current, verifiable evidence. Retrieval-augmented approaches partially address this gap but lack mechanisms to iteratively refine poor queries, whereas self-reflection methods kick in only after full retrieval is completed. In this context, we introduce PubMed Reasoner, a biomedical QA agent composed of three stages: self-critic query refinement evaluates MeSH terms for coverage, alignment, and redundancy to enhance PubMed queries based on partial (metadata) retrieval; reflective retrieval processes articles in batches until sufficient evidence is gathered; and evidence-grounded response generation produces answers with explicit citations. PubMed Reasoner with a GPT-4o backbone achieves 78.32% accuracy on PubMedQA, slightly surpassing human experts, and showing consistent gains on MMLU Clinical Knowledge. Moreover, LLM-as-judge evaluations prefer our responses across: reasoning soundness, evidence grounding, clinical relevance, and trustworthiness. By orchestrating retrieval-first reasoning over authoritative sources, our approach provides practical assistance to clinicians and biomedical researchers while controlling compute and token costs.

cs.CL↗

arXiv2026

Who Shapes Brazil's Vaccine Debate? Semi-Supervised Modeling of Stance and Polarization in YouTube's Media Ecosystem

Vaccination remains a cornerstone of global public health, yet the COVID-19 pandemic exposed how online misinformation, political polarization, and declining institutional trust can undermine immunization efforts. Most of the prior computational studies that analyzed vaccine discourse on social platforms focus on English-language data, specific vaccines, or short time windows, impairing our understanding of long-term dynamics in high-impact, non-English contexts like Brazil, home to one of the world's most comprehensive immunization systems. We here present the largest longitudinal study of Brazil's vaccine discourse on YouTube, leveraging a semi-supervised stance detection framework that combines self-labeling and self-training to classify nearly 1.4 million comments. By integrating stance with temporal patterns, engagement metrics, and channel taxonomy (legacy media, science communicators, digital-native outlets), we map how pro- and anti-vaccine narratives evolve and circulate within a hybrid media ecosystem. Our results show that semi-supervised learning substantially improves stance classification robustness, enabling fine-grained tracking of public attitudes across Brazil's full immunization schedule. Polarization spikes during epidemiological crises, especially COVID-19, but becomes fragmented across vaccines and interaction patterns in the post-pandemic period. Notably, science communication and digital-native channels emerge as the primary loci of both supportive and oppositional engagement, revealing structural vulnerabilities in contemporary health communication. Thus, our work advances computational methods for large-scale stance modeling while offering actionable evidence for public health agencies, platform governance, and online information ecosystems.

cs.CY↗

arXiv2025

AI-Powered Detection of Inappropriate Language in Medical School Curricula

The use of inappropriate language -- such as outdated, exclusionary, or non-patient-centered terms -- medical instructional materials can significantly influence clinical training, patient interactions, and health outcomes. Despite their reputability, many materials developed over past decades contain examples now considered inappropriate by current medical standards. Given the volume of curricular content, manually identifying instances of inappropriate use of language (IUL) and its subcategories for systematic review is prohibitively costly and impractical. To address this challenge, we conduct a first-in-class evaluation of small language models (SLMs) fine-tuned on labeled data and pre-trained LLMs with in-context learning on a dataset containing approximately 500 documents and over 12,000 pages. For SLMs, we consider: (1) a general IUL classifier, (2) subcategory-specific binary classifiers, (3) a multilabel classifier, and (4) a two-stage hierarchical pipeline for general IUL detection followed by multilabel classification. For LLMs, we consider variations of prompts that include subcategory definitions and/or shots. We found that both LLama-3 8B and 70B, even with carefully curated shots, are largely outperformed by SLMs. While the multilabel classifier performs best on annotated data, supplementing training with unflagged excerpts as negative examples boosts the specific classifiers' AUC by up to 25%, making them most effective models for mitigating harmful language in medical curricula.

cs.CL↗

arXiv2025

Reducing Biases towards Minoritized Populations in Medical Curricular Content via Artificial Intelligence for Fairer Health Outcomes

Biased information (recently termed bisinformation) continues to be taught in medical curricula, often long after having been debunked. In this paper, we introduce BRICC, a firstin-class initiative that seeks to mitigate medical bisinformation using machine learning to systematically identify and flag text with potential biases, for subsequent review in an expert-in-the-loop fashion, thus greatly accelerating an otherwise labor-intensive process. A gold-standard BRICC dataset was developed throughout several years, and contains over 12K pages of instructional materials. Medical experts meticulously annotated these documents for bias according to comprehensive coding guidelines, emphasizing gender, sex, age, geography, ethnicity, and race. Using this labeled dataset, we trained, validated, and tested medical bias classifiers. We test three classifier approaches: a binary type-specific classifier, a general bias classifier; an ensemble combining bias type-specific classifiers independently-trained; and a multitask learning (MTL) model tasked with predicting both general and type-specific biases. While MTL led to some improvement on race bias detection in terms of F1-score, it did not outperform binary classifiers trained specifically on each task. On general bias detection, the binary classifier achieves up to 0.923 of AUC, a 27.8% improvement over the baseline. This work lays the foundations for debiasing medical curricula by exploring a novel dataset and evaluating different training model strategies. Hence, it offers new pathways for more nuanced and effective mitigation of bisinformation.

cs.CY↗

arXiv2025

Stop treating `AGI' as the north-star goal of AI research

The AI research community plays a vital role in shaping the scientific, engineering, and societal goals of AI research. In this position paper, we argue that focusing on the highly contested topic of `artificial general intelligence' (`AGI') undermines our ability to choose effective goals. We identify six key traps -- obstacles to productive goal setting -- that are aggravated by AGI discourse: Illusion of Consensus, Supercharging Bad Science, Presuming Value-Neutrality, Goal Lottery, Generality Debt, and Normalized Exclusion. To avoid these traps, we argue that the AI research community needs to (1) prioritize specificity in engineering and societal goals, (2) center pluralism about multiple worthwhile approaches to multiple valuable goals, and (3) foster innovation through greater inclusion of disciplines and communities. Therefore, the AI research community needs to stop treating `AGI' as the north-star goal of AI research.

cs.CY↗

arXiv2025

CLaDMoP: Learning Transferrable Models from Successful Clinical Trials via LLMs

Many existing models for clinical trial outcome prediction are optimized using task-specific loss functions on trial phase-specific data. While this scheme may boost prediction for common diseases and drugs, it can hinder learning of generalizable representations, leading to more false positives/negatives. To address this limitation, we introduce CLaDMoP, a new pre-training approach for clinical trial outcome prediction, alongside the Successful Clinical Trials dataset(SCT), specifically designed for this task. CLaDMoP leverages a Large Language Model-to encode trials' eligibility criteria-linked to a lightweight Drug-Molecule branch through a novel multi-level fusion technique. To efficiently fuse long embeddings across levels, we incorporate a grouping block, drastically reducing computational overhead. CLaDMoP avoids reliance on task-specific objectives by pre-training on a "pair matching" proxy task. Compared to established zero-shot and few-shot baselines, our method significantly improves both PR-AUC and ROC-AUC, especially for phase I and phase II trials. We further evaluate and perform ablation on CLaDMoP after Parameter-Efficient Fine-Tuning, comparing it to state-of-the-art supervised baselines, including MEXA-CTP, on the Trial Outcome Prediction(TOP) benchmark. CLaDMoP achieves up to 10.5% improvement in PR-AUC and 3.6% in ROC-AUC, while attaining comparable F1 score to MEXA-CTP, highlighting its potential for clinical trial outcome prediction. Code and SCT dataset can be downloaded from https://github.com/murai-lab/CLaDMoP.

cs.LG↗

arXiv2025

KEDRec-LM: A Knowledge-distilled Explainable Drug Recommendation Large Language Model

Drug discovery is a critical task in biomedical natural language processing (NLP), yet explainable drug discovery remains underexplored. Meanwhile, large language models (LLMs) have shown remarkable abilities in natural language understanding and generation. Leveraging LLMs for explainable drug discovery has the potential to improve downstream tasks and real-world applications. In this study, we utilize open-source drug knowledge graphs, clinical trial data, and PubMed publications to construct a comprehensive dataset for the explainable drug discovery task, named \textbf{expRxRec}. Furthermore, we introduce \textbf{KEDRec-LM}, an instruction-tuned LLM which distills knowledge from rich medical knowledge corpus for drug recommendation and rationale generation. To encourage further research in this area, we will publicly release\footnote{A copy is attached with this submission} both the dataset and KEDRec-LM.

cs.CL↗

arXiv2025

MEXA-CTP: Mode Experts Cross-Attention for Clinical Trial Outcome Prediction

Clinical trials are the gold standard for assessing the effectiveness and safety of drugs for treating diseases. Given the vast design space of drug molecules, elevated financial cost, and multi-year timeline of these trials, research on clinical trial outcome prediction has gained immense traction. Accurate predictions must leverage data of diverse modes such as drug molecules, target diseases, and eligibility criteria to infer successes and failures. Previous Deep Learning approaches for this task, such as HINT, often require wet lab data from synthesized molecules and/or rely on prior knowledge to encode interactions as part of the model architecture. To address these limitations, we propose a light-weight attention-based model, MEXA-CTP, to integrate readily-available multi-modal data and generate effective representations via specialized modules dubbed "mode experts", while avoiding human biases in model design. We optimize MEXA-CTP with the Cauchy loss to capture relevant interactions across modes. Our experiments on the Trial Outcome Prediction (TOP) benchmark demonstrate that MEXA-CTP improves upon existing approaches by, respectively, up to 11.3% in F1 score, 12.2% in PR-AUC, and 2.5% in ROC-AUC, compared to HINT. Ablation studies are provided to quantify the effectiveness of each component in our proposed method.

cs.LG↗

arXiv2024

Towards Fairer Health Recommendations: finding informative unbiased samples via Word Sense Disambiguation

There have been growing concerns around high-stake applications that rely on models trained with biased data, which consequently produce biased predictions, often harming the most vulnerable. In particular, biased medical data could cause health-related applications and recommender systems to create outputs that jeopardize patient care and widen disparities in health outcomes. A recent framework titled Fairness via AI posits that, instead of attempting to correct model biases, researchers must focus on their root causes by using AI to debias data. Inspired by this framework, we tackle bias detection in medical curricula using NLP models, including LLMs, and evaluate them on a gold standard dataset containing 4,105 excerpts annotated by medical experts for bias from a large corpus. We build on previous work by coauthors which augments the set of negative samples with non-annotated text containing social identifier terms. However, some of these terms, especially those related to race and ethnicity, can carry different meanings (e.g., "white matter of spinal cord"). To address this issue, we propose the use of Word Sense Disambiguation models to refine dataset quality by removing irrelevant sentences. We then evaluate fine-tuned variations of BERT models as well as GPT models with zero- and few-shot prompting. We found LLMs, considered SOTA on many NLP tasks, unsuitable for bias detection, while fine-tuned BERT models generally perform well across all evaluated metrics.

cs.CL↗

arXiv2024

Hidden or Inferred: Fair Learning-To-Rank with Unknown Demographics

As learning-to-rank models are increasingly deployed for decision-making in areas with profound life implications, the FairML community has been developing fair learning-to-rank (LTR) models. These models rely on the availability of sensitive demographic features such as race or sex. However, in practice, regulatory obstacles and privacy concerns protect this data from collection and use. As a result, practitioners may either need to promote fairness despite the absence of these features or turn to demographic inference tools to attempt to infer them. Given that these tools are fallible, this paper aims to further understand how errors in demographic inference impact the fairness performance of popular fair LTR strategies. In which cases would it be better to keep such demographic attributes hidden from models versus infer them? We examine a spectrum of fair LTR strategies ranging from fair LTR with and without demographic features hidden versus inferred to fairness-unaware LTR followed by fair re-ranking. We conduct a controlled empirical investigation modeling different levels of inference errors by systematically perturbing the inferred sensitive attribute. We also perform three case studies with real-world datasets and popular open-source inference methods. Our findings reveal that as inference noise grows, LTR-based methods that incorporate fairness considerations into the learning process may increase bias. In contrast, fair re-ranking strategies are more robust to inference errors. All source code, data, and experimental artifacts of our experimental study are available here: https://github.com/sewen007/hoiltr.git

cs.LG↗

arXiv2024

General Performance Evaluation for Competitive Resource Allocation Games via Unseen Payoff Estimation

Many high-stakes decision-making problems, such as those found within cybersecurity and economics, can be modeled as competitive resource allocation games. In these games, multiple players must allocate limited resources to overcome their opponent(s), while minimizing any induced individual losses. However, existing means of assessing the performance of resource allocation algorithms are highly disparate and problem-dependent. As a result, evaluating such algorithms is unreliable or impossible in many contexts and applications, especially when considering differing levels of feedback. To resolve this problem, we propose a generalized definition of payoff which uses an arbitrary user-provided function. This unifies performance evaluation under all contexts and levels of feedback. Using this definition, we develop metrics for evaluating player performance, and estimators to approximate them under uncertainty (i.e., bandit or semi-bandit feedback). These metrics and their respective estimators provide a problem-agnostic means to contextualize and evaluate algorithm performance. To validate the accuracy of our estimator, we explore the Colonel Blotto ($\mathcal{CB}$) game as an example. To this end, we propose a graph-pruning approach to efficiently identify feasible opponent decisions, which are used in computing our estimation metrics. Using various resource allocation algorithms and game parameters, a suite of $\mathcal{CB}$ games are simulated and used to compute and evaluate the quality of our estimates. These simulations empirically show our approach to be highly accurate at estimating the metrics associated with the unseen outcomes of an opponent's latent behavior.

cs.GT↗

arXiv2023

Towards Detecting Cascades of Biased Medical Claims on Twitter

Social media may disseminate medical claims that highlight misleading correlations between social identifiers and diseases due to not accounting for structural determinants of health. Our research aims to identify biased medical claims on Twitter and measure their spread. We propose a machine learning framework that uses two models in tandem: RoBERTa to detect medical claims and DistilBERT to classify bias. After identifying original biased medical claims, we conducted a retweet cascade analysis, computing their individual reach and rate of spread. Tweets containing biased claims were found to circulate faster and further than unbiased claims.

cs.SI↗

arXiv2023

Helping Fact-Checkers Identify Fake News Stories Shared through Images on WhatsApp

WhatsApp has introduced a novel avenue for smartphone users to engage with and disseminate news stories. The convenience of forming interest-based groups and seamlessly sharing content has rendered WhatsApp susceptible to the exploitation of misinformation campaigns. While the process of fact-checking remains a potent tool in identifying fabricated news, its efficacy falters in the face of the unprecedented deluge of information generated on the Internet today. In this work, we explore automatic ranking-based strategies to propose a "fakeness score" model as a means to help fact-checking agencies identify fake news stories shared through images on WhatsApp. Based on the results, we design a tool and integrate it into a real system that has been used extensively for monitoring content during the 2018 Brazilian general election. Our experimental evaluation shows that this tool can reduce by up to 40% the amount of effort required to identify 80% of the fake news in the data when compared to current mechanisms practiced by the fact-checking agencies for the selection of news stories to be checked.

cs.CY↗

arXiv2022

DELATOR: Money Laundering Detection via Multi-Task Learning on Large Transaction Graphs

Money laundering has become one of the most relevant criminal activities in modern societies, as it causes massive financial losses for governments, banks and other institutions. Detecting such activities is among the top priorities when it comes to financial analysis, but current approaches are often costly and labor intensive partly due to the sheer amount of data to be analyzed. Hence, there is a growing need for automatic anti-money laundering systems to assist experts. In this work, we propose DELATOR, a novel framework for detecting money laundering activities based on graph neural networks that learn from large-scale temporal graphs. DELATOR provides an effective and efficient method for learning from heavily imbalanced graph data, by adapting concepts from the GraphSMOTE framework and incorporating elements of multi-task learning to obtain rich node embeddings for node classification. DELATOR outperforms all considered baselines, including an off-the-shelf solution from Amazon AWS by 23% with respect to AUC-ROC. We also conducted real experiments that led to the discovery of 7 new suspicious cases among the 50 analyzed ones, which have been reported to the authorities.

cs.LG↗

arXiv2022

On the Dynamics of Political Discussions on Instagram: A Network Perspective

Instagram has been increasingly used as a source of information especially among the youth. As a result, political figures now leverage the platform to spread opinions and political agenda. We here analyze online discussions on Instagram, notably in political topics, from a network perspective. Specifically, we investigate the emergence of communities of co-commenters, that is, groups of users who often interact by commenting on the same posts and may be driving the ongoing online discussions. In particular, we are interested in salient co-interactions, i.e., interactions of co-commenters that occur more often than expected by chance and under independent behavior. Unlike casual and accidental co-interactions which normally happen in large volumes, salient co-interactions are key elements driving the online discussions and, ultimately, the information dissemination. We base our study on the analysis of 10 weeks of data centered around major elections in Brazil and Italy, following both politicians and other celebrities. We extract and characterize the communities of co-commenters in terms of topological structure, properties of the discussions carried out by community members, and how some community properties, notably community membership and topics, evolve over time. We show that communities discussing political topics tend to be more engaged in the debate by writing longer comments, using more emojis, hashtags and negative words than in other subjects. Also, communities built around political discussions tend to be more dynamic, although top commenters remain active and preserve community membership over time. Moreover, we observe a great diversity in discussed topics over time: whereas some topics attract attention only momentarily, others, centered around more fundamental political discussions, remain consistently active over time.

cs.SI↗

arXiv2021

Top-Down Deep Clustering with Multi-generator GANs

Deep clustering (DC) leverages the representation power of deep architectures to learn embedding spaces that are optimal for cluster analysis. This approach filters out low-level information irrelevant for clustering and has proven remarkably successful for high dimensional data spaces. Some DC methods employ Generative Adversarial Networks (GANs), motivated by the powerful latent representations these models are able to learn implicitly. In this work, we propose HC-MGAN, a new technique based on GANs with multiple generators (MGANs), which have not been explored for clustering. Our method is inspired by the observation that each generator of a MGAN tends to generate data that correlates with a sub-region of the real data distribution. We use this clustered generation to train a classifier for inferring from which generator a given image came from, thus providing a semantically meaningful clustering for the real distribution. Additionally, we design our method so that it is performed in a top-down hierarchical clustering tree, thus proposing the first hierarchical DC method, to the best of our knowledge. We conduct several experiments to evaluate the proposed method against recent DC methods, obtaining competitive results. Last, we perform an exploratory analysis of the hierarchical clustering tree that highlights how accurately it organizes the data in a hierarchy of semantically coherent patterns.

cs.LG↗