arXiv subjects

Bernardo Modenesi

Publications and source records attributed to Bernardo Modenesi.

11 recordsLinked to original sources

arXiv2026

Prompting is not enough: supervised baselines and leakage control for measuring shared decision-making with LLMs in pediatric encounters

Objectives: To determine whether zero-shot prompting of a large language model (LLM) is sufficient to detect shared decision-making (SDM) behaviors in real clinical encounters, and whether supervised learning adds value under patient-grouped, nested evaluation. Methods: We analyzed 21 audio-recorded outpatient surgical decision encounters (19 unique patients; 7,566 utterance segments; ~6.1 hours) between families of children with multiple long-term conditions and their surgical providers. Trained coders labeled segments for 12 SDM behaviors (human-human macro Cohen's kappa = 0.695). We compared a zero-shot local LLM (Qwen 2.5 32B), a supervised classifier over frozen sentence embeddings, and their logistic stack, under patient-grouped outer folds with inner cross-fitted thresholds and patient-resampled confidence intervals. Results: The zero-shot LLM reached macro kappa = 0.139 (95% CI 0.111-0.164). The supervised classifier reached kappa = 0.227 (0.186-0.262), a paired improvement of 0.088 (0.051-0.119). A logistic stack of the two reached kappa = 0.242 (0.198-0.284). We identified multiple corpus-specific leakage paths, including grouping sibling recordings separately and allowing labels from an outer held-out patient to enter few-shot exemplars used while fitting downstream models. Conclusion: Zero-shot prompting alone is not sufficient to measure SDM behavior as reliably as a small supervised model, and patient-level grouping alone does not prevent leakage when labeled prompt exemplars are precomputed outside the outer evaluation loop. Reported performance is sensitive to the unit of data splitting and to where labeled exemplars enter the pipeline. External validation is needed before these findings generalize beyond this population, model, prompt, and codebook.

cs.CL↗

arXiv2026

When Should Graph Attention Be Sparse? Learning a Per-Edge Tsallis Index

Graph attention normalizes neighborhood scores with softmax, the maximum-entropy choice under Shannon statistics. But homophilic and heterophilic graphs want different attention shapes, and one fixed normalization cannot serve both. We propose \textbf{LTGA} (\textbf{L}earnable \textbf{T}sallis \textbf{G}raph \textbf{A}ttention), a graph attention layer whose Tsallis entropic index $q$ is learned jointly with the weights, interpolating continuously between heavy-tailed ($q\!<\!1$), softmax ($q\!=\!1$) and compact-support ($q\!>\!1$) attention at four granularities from a global scalar to a per-edge index, under a bounded reparameterization that starts every model at the GAT baseline. Across eight benchmarks at ten seeds, LTGA-Edge takes the best average rank ($2.75$), but the omnibus test does not reject ($p\!=\!0.199$) and learning $q$ does not beat searching it: a validation-tuned frozen grid reaches $61.4\%$, tuned $\alpha$-entmax $62.2\%$ and a capacity-matched $q\!\equiv\!1$ control $62.0\%$, against $61.7\%$ for LTGA-Edge. What the learned index buys is one run instead of a grid, and an interpretable mechanism: where $q$ leaves $1$, it prunes $42\%$ of attention coefficients to exactly zero, and those edges are selectively the wrong ones, restoring them costs $7.1$ points, while random pruning at the same rate costs $13.0$ more. Project page: https://kleyt0n.github.io/ltga

cs.LG↗

arXiv2026

Perspectives on Tsallis Statistics for Artificial Intelligence

Tsallis statistics generalizes Boltzmann-Gibbs statistical mechanics through a single real parameter $q$ that controls the weight assigned to rare and frequent events. Originally proposed to describe physical systems with long-range correlations, multifractal geometry, and heavy-tailed fluctuations, the framework has become a recurring ingredient in modern artificial intelligence (AI): it underlies sparse attention mechanisms (\textsc{sparsemax} and $\alpha$-\textsc{entmax}), maximum-entropy reinforcement learning with controllable exploration, robust and heavy-tailed probabilistic models, and a family of generalized loss functions and regularizers. This paper offers a structured perspective on where Tsallis statistics meets AI. We first review the mathematical core: $q$-entropy and its variational (maximum-entropy) foundation, the $q$-exponential and $q$-logarithm, the $q$-central limit theorem, $q$-Gaussian distributions, and their dynamical origin in superstatistics, emphasizing the properties that matter for machine learning. We then survey applications across softmax generalization, reinforcement learning, sequential and graph neural models, generative and probabilistic modeling, loss design, and optimization, extracting the recurring design pattern in each case: a tunable interpolation between dense/uniform and sparse/peaked behavior governed by $q$. We further argue that the heavy-tailed weight spectra and gradient-noise statistics empirically observed in deep networks are themselves nonextensive signatures, placing modern learning dynamics within the scope of $q$-statistics. Finally, we discuss methodological pitfalls, the relationship to information geometry and $q$-exponential families, and open directions, arguing that $q$ should be treated as a learnable inductive bias rather than a fixed hyperparameter.

cs.AI↗

arXiv2026

GraphNetz: Statistical Benchmarking of Graph Neural Networks with Paired Tests and Rank Aggregation

Graph Neural Networks (GNNs) benchmarks often report single point estimates, even when performance differences are small relative to variation across random seeds, train/test splits, and datasets. Confidence intervals, paired comparisons, multiple-comparison correction, and rank-based aggregation are standard statistical tools, but they are rarely the default output of graph-learning benchmark suites. We introduce GraphNetz, a benchmarking framework whose default output is a structured statistical report rather than a raw accuracy table. GraphNetz currently includes 63 dataset loaders, four task types, and five canonical GNN architectures, while also supporting custom datasets and models. The framework standardizes multi-seed evaluation and automatically returns per-cell confidence intervals, Holm-corrected paired tests, and Friedman-Nemenyi critical-difference diagrams across tasks. In a cross-category benchmark over ten heterogeneous tasks, apparent rank differences among four canonical node-level encoders fall within a single Nemenyi clique, indicating that none is significantly better than the others at $\alpha = 0.05$. GraphNetz therefore provides researchers with a reproducible computational and statistical pipeline to benchmark new graph-learning methods against standard architectures, over different tasks and a wide set of applications, while reporting principled statistical evidence for benchmarking which accounts for seed uncertainty. This framework is set to serve the graph-learning community with a reproducible and honest model comparison ready to be added to papers.

cs.CE↗

arXiv2026

Divergence-Guided Particle Swarm Optimization

Particle Swarm Optimization (PSO) is susceptible to premature convergence when the swarm collapses around the global best, particularly on multimodal landscapes in higher dimensions. We propose Divergence-guided PSO (DPSO), which augments the velocity update with a modulation term that repels particles whose personal bests have converged near the global best. The repulsion is gated by a Gaussian similarity kernel, which we prove is equivalent to an exponentially decaying function of the KL divergence between Gaussian-embedded personal and global bests, connecting the mechanism to the family of $f$-divergences and providing a principled basis for kernel design. Experiments on 36 benchmark functions (15 unimodal, 21 multimodal) across dimensions $D \in \{10, 30, 50\}$, each with 30 independent runs, show that DPSO frequently outperforms standard PSO on multimodal problems, with improvements of 2-8$\times$ on functions such as Pinter, Ackley, and Levy, and up to 5$\times$ reduction in run-to-run variance. On unimodal landscapes the modulation term is counterproductive, confirming that DPSO targets the exploration-exploitation trade-off rather than offering a universal improvement. The method adds one hyperparameter, incurs 15--25\% wall-clock overhead, and does not increase the asymptotic per-iteration complexity of PSO. The project code is available here: https://github.com/Kleyt0n/dpso

cs.CE↗

arXiv2025

Generalized Machine Learning for Fast Calibration of Agent-Based Epidemic Models

Agent-based models (ABMs) are widely used to study infectious disease dynamics, but their calibration is often computationally intensive, limiting their applicability in time-sensitive public health settings. We propose DeepIMC (Deep Inverse Mapping Calibration), a machine learning-based calibration framework that directly learns the inverse mapping from epidemic time series to epidemiological parameters. DeepIMC trains a bidirectional Long Short-Term Memory (BiLSTM) neural network on synthetic epidemic trajectories generated from agent-based models such as the Susceptible-Infected-Recovered (SIR) model, enabling rapid parameter estimation without repeated simulation at inference time. We evaluate DeepIMC through an extensive simulation study comprising 5,000 heterogeneous epidemic scenarios and benchmark its performance against Approximate Bayesian Computation (ABC) using likelihood-free Markov Chain Monte Carlo. The results show that DeepIMC substantially improves parameter recovery accuracy, produces sharp and well-calibrated predictive intervals, and reduces computational time by more than an order of magnitude relative to ABC. Although structural parameter identifiability constraints limit the precise recovery of all model parameters simultaneously, the calibrated models reliably reproduce epidemic trajectories and support accurate forward prediction with their estimated parameters. DeepIMC is implemented in the open-source R package epiworldRCalibrate, facilitating practical adoption for real-time epidemic modeling and policy analysis. Overall, our findings demonstrate that DeepIMC provides a scalable, operationally effective alternative to traditional simulation-based calibration methods for agent-based epidemic models.

cs.LG↗

arXiv2024

Advancing Distribution Decomposition Methods Beyond Common Supports: Applications to Racial Wealth Disparities

I generalize state-of-the-art approaches that decompose differences in the distribution of a variable of interest between two groups into a portion explained by covariates and a residual portion. The method that I propose relaxes the overlapping supports assumption, allowing the groups being compared to not necessarily share exactly the same covariate support. I illustrate my method revisiting the black-white wealth gap in the U.S. as a function of labor income and other variables. Traditionally used decomposition methods would trim (or assign zero weight to) observations that lie outside the common covariate support region. On the other hand, by allowing all observations to contribute to the existing wealth gap, I find that otherwise trimmed observations contribute from 3% to 19% to the overall wealth gap, at different portions of the wealth distribution.

econ.EM↗

arXiv2024

Detailed Gender Wage Gap Decompositions: Controlling for Worker Unobserved Heterogeneity Using Network Theory

Recent advances in the literature of decomposition methods in economics have allowed for the identification and estimation of detailed wage gap decompositions. In this context, building reliable counterfactuals requires using tighter controls to ensure that similar workers are correctly identified by making sure that important unobserved variables such as skills are controlled for, as well as comparing only workers with similar observable characteristics. This paper contributes to the wage decomposition literature in two main ways: (i) developing an economic principled network based approach to control for unobserved worker skills heterogeneity in the presence of potential discrimination; and (ii) extending existing generic decomposition tools to accommodate for potential lack of overlapping supports in covariates between groups being compared, which is likely to be the norm in more detailed decompositions. We illustrate the methodology by decomposing the gender wage gap in Brazil.

econ.EM↗

arXiv2023

An Interdisciplinary Outlook on Large Language Models for Scientific Research

In this paper, we describe the capabilities and constraints of Large Language Models (LLMs) within disparate academic disciplines, aiming to delineate their strengths and limitations with precision. We examine how LLMs augment scientific inquiry, offering concrete examples such as accelerating literature review by summarizing vast numbers of publications, enhancing code development through automated syntax correction, and refining the scientific writing process. Simultaneously, we articulate the challenges LLMs face, including their reliance on extensive and sometimes biased datasets, and the potential ethical dilemmas stemming from their use. Our critical discussion extends to the varying impacts of LLMs across fields, from the natural sciences, where they help model complex biological sequences, to the social sciences, where they can parse large-scale qualitative data. We conclude by offering a nuanced perspective on how LLMs can be both a boon and a boundary to scientific progress.

cs.CL↗

arXiv2023

What is a Labor Market? Classifying Workers and Jobs Using Network Theory

This paper develops a new data-driven approach to characterizing latent worker skill and job task heterogeneity by applying an empirical tool from network theory to large-scale Brazilian administrative data on worker--job matching. We microfound this tool using a standard equilibrium model of workers matching with jobs according to comparative advantage. Our classifications identify important dimensions of worker and job heterogeneity that standard classifications based on occupations and sectors miss. The equilibrium model based on our classifications more accurately predicts wage changes in response to the 2016 Olympics than a model based on occupations and sectors. Additionally, for a large simulated shock to demand for workers, we show that reduced form estimates of the effects of labor market shock exposure on workers' earnings are nearly 4 times larger when workers and jobs are classified using our classifications as opposed to occupations and sectors.

econ.GN↗

arXiv2023

Evaluating Explainability in Machine Learning Predictions through Explainer-Agnostic Metrics

The rapid integration of artificial intelligence (AI) into various industries has introduced new challenges in governance and regulation, particularly regarding the understanding of complex AI systems. A critical demand from decision-makers is the ability to explain the results of machine learning models, which is essential for fostering trust and ensuring ethical AI practices. In this paper, we develop six distinct model-agnostic metrics designed to quantify the extent to which model predictions can be explained. These metrics measure different aspects of model explainability, ranging from local importance, global importance, and surrogate predictions, allowing for a comprehensive evaluation of how models generate their outputs. Furthermore, by computing our metrics, we can rank models in terms of explainability criteria such as importance concentration and consistency, prediction fluctuation, and surrogate fidelity and stability, offering a valuable tool for selecting models based not only on accuracy but also on transparency. We demonstrate the practical utility of these metrics on classification and regression tasks, and integrate these metrics into an existing Python package for public use.

cs.LG↗