arXiv subjects

Nicolás Astorga

Publications and source records attributed to Nicolás Astorga.

11 recordsLinked to original sources

arXiv2026

Timely Clinical Diagnosis through Active Test Selection

There is growing interest in using machine learning (ML) to support clinical diagnosis, but most approaches rely on static, fully observed datasets and fail to reflect the sequential, resource-aware reasoning clinicians use in practice. Diagnosis remains complex and error prone, especially in high-pressure or resource-limited settings, underscoring the need for frameworks that help clinicians make timely and cost-effective decisions. We propose ACTMED (Adaptive Clinical Test selection via Model-based Experimental Design), a diagnostic framework that integrates Bayesian Experimental Design (BED) with large language models (LLMs) to better emulate real-world diagnostic reasoning. At each step, ACTMED selects the test expected to yield the greatest reduction in diagnostic uncertainty for a given patient. LLMs act as flexible simulators, generating plausible patient state distributions and supporting belief updates without requiring structured, task-specific training data. Clinicians can remain in the loop; reviewing test suggestions, interpreting intermediate outputs, and applying clinical judgment throughout. We evaluate ACTMED on real-world datasets and show it can optimize test selection to improve diagnostic accuracy, interpretability, and resource use. This represents a step toward transparent, adaptive, and clinician-aligned diagnostic systems that generalize across settings with reduced reliance on domain-specific data.

cs.AI↗

arXiv2026

Step-by-Step Optimization-like Reasoning in LLMs over Expanding Search Spaces

Verifiable reward training has improved mathematical and coding reasoning, but these domains capture only part of step-by-step decision making. Many real-world tasks require finding a high-value feasible plan among many valid alternatives. We introduce OPT*, a scalable family of optimization-style tasks for training and evaluating LLM step-by-step optimization-like reasoning along a complexity axis: each task provides a feasibility checker and evaluator, while a complexity parameter expands the search space without requiring new human labels. This motivates studying these tasks in two regimes: (i) solver-guided online policy optimization, which uses a solver as a value oracle for partial states and applies rank-based reward shaping to reinforce better next steps, and (ii) search-based offline RL when such solvers are unavailable. Theoretically, we relate success in large search spaces to the information a reasoner extracts per unit of search budget. Empirically, we ablate the ingredients that make search efficient on OPT* and show that training on OPT* improves step-by-step optimization-like reasoning.

cs.AI↗

arXiv2026

Learning Reasoning Rewards from Expert Demonstrations with Inverse Reinforcement Learning

Teaching large language models (LLMs) to reason during post-training typically relies on reinforcement learning with explicit outcome- or process-based reward functions. However, in many real-world settings, obtaining or defining such reward functions is difficult, especially for complex tasks, making learning from expert demonstrations an attractive alternative. The dominant approach, supervised fine-tuning (SFT), trains models to imitate expert reasoning traces directly, but suffers from the general limitations of off-policy learning: performance can be fragile to inference-time deviations from states explicitly covered by the demonstrations. To address this, we propose Reasoning Adversarial Inverse Reinforcement Learning (R-AIRL). Rather than imitating the expert's reasoning, R-AIRL infers the underlying process-level reward from the expert Chain-of-Thoughts. Through experiments on GSM8K, MMLU-Pro and MedReason we show that the reasoning reward function learned with R-AIRL can be effectively used throughout the training and inference pipeline: (1) to provide a training signal for post-training, outperforming SFT in most of the considered settings, (2) for inference-time reranking, improving pass@1 by up to 17.4 points, and (3) for process-level evaluation, localising reasoning failures with up to 86.1% accuracy. Overall, R-AIRL bridges imitation learning and reward-based optimisation, enabling the extraction of meaningful reasoning signals from expert thinking traces.

cs.AI↗

arXiv2026

CauSim: Scaling Causal Reasoning with Increasingly Complex Causal Simulators

Despite surpassing human performance across mathematics, coding, and other knowledge-intensive tasks, large language models (LLMs) continue to struggle with causal reasoning. A core obstacle is the target data itself: causal systems are complex and often expressed in non-executable forms, while ground-truth answers to causal queries are inherently scarce. We introduce CauSim, a framework that turns causal reasoning from a scarce-label problem into a scalable supervised one. CauSim constructs increasingly complex causal simulators: executable structural causal models (SCMs), incrementally built by LLMs, that scale to globally complex systems while maintaining verifiable answers to causal queries. CauSim operates across representations by formalizing non-executable causal knowledge into code, enabling data augmentation, and translating executable SCMs into natural language, enabling supervision in previously difficult-to-supervise representations. We structure our research into two parts: (1) how to construct increasingly complex causal simulators, and (2) a systematic study of what CauSim enables, demonstrating generalization across representations, consistent gains from curriculum scaling and data volume, LLM self-improvement through self-generated simulators, and data augmentation via formalization of existing domain knowledge.

cs.AI↗

arXiv2025

Continuously Updating Digital Twins using Large Language Models

Digital twins are models of real-world systems that can simulate their dynamics in response to potential actions. In complex settings, the state and action variables, and available data and knowledge relevant to a system can constantly change, requiring digital twins to continuously update with these changes to remain relevant. Current approaches struggle in this regard, as they require fixed, well-defined modelling environments, and they cannot adapt to novel variables without re-designs, or incorporate new information without re-training. To address this, we frame digital twinning as an in-context learning problem using large language models, enabling seamless updates to the twin at inference time. We develop CALM-DT, a Context-Adaptive Language Model-based Digital Twin that can accurately simulate across diverse state-action spaces using in-context learning alone by utilising fine-tuned encoders for sample retrieval. We empirically demonstrate CALM-DT's competitive performance with existing digital twin approaches, and its unique ability to adapt to changes in its modelling environment without parameter updates.

cs.CL↗

arXiv2025

Autoformulation of Mathematical Optimization Models Using LLMs

Mathematical optimization is fundamental to decision-making across diverse domains, from operations research to healthcare. Yet, translating real-world problems into optimization models remains a difficult task, often demanding specialized expertise. This paper approaches the problem of $\textit{autoformulation}$: the automated creation of solver-ready optimization models from natural language problem descriptions. We identify three core challenges of autoformulation: $\textit{(1)}$ the vast, problem-dependent hypothesis space, $\textit{(2)}$ efficient and diverse exploration of this space under uncertainty, and $\textit{(3)}$ evaluation of formulation correctness against problem description. To address these challenges, we present a novel method leveraging $\textit{Large Language Models}$ (LLMs) with $\textit{Monte-Carlo Tree Search}$, exploiting the hierarchical nature of optimization modeling to generate and systematically explore possible formulations. To enhance search efficiency, we introduce symbolic pruning to eliminate trivially equivalent search paths (branches), and employ LLM-based evaluation of partial formulations to guide search. Empirical analysis on linear and mixed-integer programming benchmarks demonstrates our method's effectiveness, with significant performance gains from both LLM-based value estimation and symbolic pruning techniques.

cs.LG↗

arXiv2024

Large Language Models to Enhance Bayesian Optimization

Bayesian optimization (BO) is a powerful approach for optimizing complex and expensive-to-evaluate black-box functions. Its importance is underscored in many applications, notably including hyperparameter tuning, but its efficacy depends on efficiently balancing exploration and exploitation. While there has been substantial progress in BO methods, striking this balance remains a delicate process. In this light, we present LLAMBO, a novel approach that integrates the capabilities of Large Language Models (LLM) within BO. At a high level, we frame the BO problem in natural language, enabling LLMs to iteratively propose and evaluate promising solutions conditioned on historical evaluations. More specifically, we explore how combining contextual understanding, few-shot learning proficiency, and domain knowledge of LLMs can improve model-based BO. Our findings illustrate that LLAMBO is effective at zero-shot warmstarting, and enhances surrogate modeling and candidate sampling, especially in the early stages of search when observations are sparse. Our approach is performed in context and does not require LLM finetuning. Additionally, it is modular by design, allowing individual components to be integrated into existing BO frameworks, or function cohesively as an end-to-end method. We empirically validate LLAMBO's efficacy on the problem of hyperparameter tuning, highlighting strong empirical performance across a range of diverse benchmarks, proprietary, and synthetic tasks.

cs.LG↗

arXiv2023

Multi-Class Deep SVDD: Anomaly Detection Approach in Astronomy with Distinct Inlier Categories

With the increasing volume of astronomical data generated by modern survey telescopes, automated pipelines and machine learning techniques have become crucial for analyzing and extracting knowledge from these datasets. Anomaly detection, i.e. the task of identifying irregular or unexpected patterns in the data, is a complex challenge in astronomy. In this paper, we propose Multi-Class Deep Support Vector Data Description (MCDSVDD), an extension of the state-of-the-art anomaly detection algorithm One-Class Deep SVDD, specifically designed to handle different inlier categories with distinct data distributions. MCDSVDD uses a neural network to map the data into hyperspheres, where each hypersphere represents a specific inlier category. The distance of each sample from the centers of these hyperspheres determines the anomaly score. We evaluate the effectiveness of MCDSVDD by comparing its performance with several anomaly detection algorithms on a large dataset of astronomical light-curves obtained from the Zwicky Transient Facility. Our results demonstrate the efficacy of MCDSVDD in detecting anomalous sources while leveraging the presence of different inlier categories. The code and the data needed to reproduce our results are publicly available at https://github.com/mperezcarrasco/AnomalyALeRCE.

cs.LG↗

arXiv2023

Multi-scale stamps for real-time classification of alert streams

In recent years, automatic classifiers of image cutouts (also called "stamps") have shown to be key for fast supernova discovery. The Vera C. Rubin Observatory will distribute about ten million alerts with their respective stamps each night, enabling the discovery of approximately one million supernovae each year. A growing source of confusion for these classifiers is the presence of satellite glints, sequences of point-like sources produced by rotating satellites or debris. The currently planned Rubin stamps will have a size smaller than the typical separation between these point sources. Thus, a larger field of view stamp could enable the automatic identification of these sources. However, the distribution of larger stamps would be limited by network bandwidth restrictions. We evaluate the impact of using image stamps of different angular sizes and resolutions for the fast classification of events (AGNs, asteroids, bogus, satellites, SNe, and variable stars), using data from the Zwicky Transient Facility. We compare four scenarios: three with the same number of pixels (small field of view with high resolution, large field of view with low resolution, and a multi-scale proposal) and a scenario with the full stamp that has a larger field of view and higher resolution. Compared to small field of view stamps, our multi-scale strategy reduces misclassifications of satellites as asteroids or supernovae, performing on par with high-resolution stamps that are 15 times heavier. We encourage Rubin and its Science Collaborations to consider the benefits of implementing multi-scale stamps as a possible update to the alert specification.

astro-ph.IM↗

arXiv2021

Matching Embeddings for Domain Adaptation

In this work we address the problem of transferring knowledge obtained from a vast annotated source domain to a low labeled target domain. We propose Adversarial Variational Domain Adaptation (AVDA), a semi-supervised domain adaptation method based on deep variational embedded representations. We use approximate inference and domain adversarial methods to map samples from source and target domains into an aligned class-dependent embedding defined as a Gaussian Mixture Model. AVDA works as a classifier and considers a generative model that helps this classification. We used digits dataset for experimentation. Our results show that on a semi-supervised few-shot scenario our model outperforms previous methods in most of the adaptation tasks, even using a fewer number of labeled samples per class on target domain.

cs.LG↗

arXiv2020

MPCC: Matching Priors and Conditionals for Clustering

Clustering is a fundamental task in unsupervised learning that depends heavily on the data representation that is used. Deep generative models have appeared as a promising tool to learn informative low-dimensional data representations. We propose Matching Priors and Conditionals for Clustering (MPCC), a GAN-based model with an encoder to infer latent variables and cluster categories from data, and a flexible decoder to generate samples from a conditional latent space. With MPCC we demonstrate that a deep generative model can be competitive/superior against discriminative methods in clustering tasks surpassing the state of the art over a diverse set of benchmark datasets. Our experiments show that adding a learnable prior and augmenting the number of encoder updates improve the quality of the generated samples, obtaining an inception score of 9.49 $\pm$ 0.15 and improving the Fréchet inception distance over the state of the art by a 46.9% in CIFAR10.

cs.LG↗