arXiv subjects

Stefan T. Radev

Publications and source records attributed to Stefan T. Radev.

At least 19 recordsLinked to original sources

arXiv2026

Diffusion Models in Simulation-Based Inference: A Tutorial Review

Diffusion models have recently emerged as powerful learners for simulation-based inference (SBI), enabling fast and accurate estimation of latent parameters from simulated and real data. Their score-based formulation offers a flexible way to learn conditional or joint distributions over parameters and observations, thereby providing a versatile solution to various modeling problems. In this tutorial review, we synthesize recent developments on diffusion models for SBI, covering design choices for training, inference, and evaluation. We highlight opportunities created by various concepts such as guidance, score composition, flow matching, consistency models, and joint modeling. Furthermore, we discuss how efficiency and statistical accuracy are affected by noise schedules, parameterizations, and samplers. Finally, we illustrate these concepts with case studies across parameter dimensionalities, simulation budgets, and model types, and outline open questions for future research.

stat.ML↗

arXiv2026

Inverting Foundation Models of Brain Function with Simulation-Based Inference

Foundation models of brain activity promise a new frontier for in silico neuroscience by emulating neural responses to complex stimuli across tasks and modalities. A natural next step is to ask whether these models can also be used in reverse. Can we recover a stimulus or its properties from synthetic brain activity? We study this question in a proof-of-concept setting using TRIBEv2. We pair the brain emulator with large language models (LLMs) that generate news headlines from linguistic parameters such as valence, arousal, and dominance. We then use simulation-based inference to learn a probabilistic mapping from brain maps to latent stimulus parameters. Our results show that these parameters can be recovered from predicted brain maps, demonstrating that the emulator's synthetic neural encodings preserve information about the controlled stimulus dimensions. They also show that LLMs can serve as controllable stimulus generators for simulated experiments. Together, these findings provide a step toward decoding and inverse design with foundation brain models.

cs.LG↗

arXiv2026

JADAI: Jointly Amortizing Adaptive Design and Bayesian Inference

We consider problems of parameter estimation where design variables can be actively optimized to maximize information gain. To this end, we introduce JADAI, a framework that jointly amortizes Bayesian adaptive design and inference by training a policy, a history network, and an inference network end-to-end. The networks minimize a generic loss that aggregates incremental reductions in posterior error along experimental sequences. Inference networks are instantiated with diffusion-based posterior estimators that can approximate high-dimensional and multimodal posteriors at every experimental step. Across standard adaptive design benchmarks, JADAI achieves superior or competitive performance.

stat.ML↗

arXiv2026

CogFormer: Learn All Your Models Once

Simulation-based inference (SBI) with neural networks has accelerated and transformed cognitive modeling workflows. SBI enables modelers to fit complex models that were previously difficult or impossible to estimate, while also allowing rapid estimation across large numbers of datasets. However, the utility of SBI for iterating over varying modeling assumptions remains limited: changes to parameterizations, generative functions, priors, and design variables all necessitate model retraining, thereby diminishing the benefits of amortization. To address these issues, we pilot the CogFormer, a meta-amortized framework for cognitive modeling. Our framework trains a transformer-based architecture that remains valid across a combinatorial number of structurally similar models, allowing for changing data types, parameters, design matrices, and sample sizes. We present promising quantitative results across families of decision-making models for binary, multi-alternative, and continuous responses. Our evaluation suggests that CogFormer can accurately estimate parameters across model families with minimal amortization offset, making it a potentially powerful engine that catalyzes cognitive modeling workflows.

stat.ML↗

arXiv2026

High-Fidelity 3D Simulator for Synthetic fNIRS Data Generation

Functional near-infrared spectroscopy (fNIRS) provides a noninvasive window into brain activity by measuring task-related changes in oxygenated and deoxygenated hemoglobin in the cortex. A key advantage of fNIRS is its promise of use with mobile participants in complex, real-world environments, such as walking, sports, classroom learning, driving simulations, or social interactions. However, analyzing fNIRS data is challenging because of motion artifacts, physiological noise, and other confounding factors. This challenge is further compounded by the limited availability of annotated datasets, which hinders the development and validation of new analysis pipelines, particularly given the growing use of AI methods. Recognizing these challenges, we introduce a 3D fNIRS simulator that uses mesh-based Monte Carlo simulations to create physiologically realistic, full-head synthetic recordings with high spatiotemporal fidelity. Our simulator combines anatomically accurate sensitivity profiles with parameterized models of hemodynamic responses, systemic physiology, and nonsystematic artifacts. As a result, users can generate virtually unlimited labeled datasets for testing denoising algorithms, data augmentation, mechanistic modeling, or \textit{in silico} experimentation. We validate the simulator using experimental fNIRS data from open-source finger-tapping, pain-assessment, and surgical-skill datasets and provide an open-source implementation to support reproducibility and broad adoption.

q-bio.NC↗

arXiv2026

Improving the Accuracy of Amortized Model Comparison with Self-Consistency

Amortized Bayesian model comparison (BMC) enables fast probabilistic ranking of models via simulation-based training of neural surrogates. However, the accuracy of neural surrogates deteriorates when simulation models are misspecified; the very case where model comparison is most needed. We evaluate four different amortized BMC methods. We supplement traditional simulation-based training of these methods with a \emph{self-consistency} (SC) loss on unlabeled real data to improve BMC estimates under distribution shifts. Using one artificial and two real-world case studies, we compare amortized BMC estimators with and without SC against analytic or bridge sampling benchmarks. In the \emph{closed-world} case (data is generated by one of the candidate models), BMC estimators using classifiers work acceptably well even without SC training. However, these methods also benefit the least from SC training. In the \emph{open-world} scenario (all models misspecified), SC training strongly improves BMC estimators when having access to analytic likelihoods, or when surrogate likelihoods are locally accurate near the true parameter posterior, even for severely misspecified models. We conclude with practical recommendations for amortized BMC and suggestions for future research.

stat.ML↗

arXiv2026

Compositional amortized inference for large-scale hierarchical Bayesian models

Amortized Bayesian inference (ABI) with neural networks has emerged as a powerful simulation-based approach for estimating complex mechanistic models. However, extending ABI to hierarchical models, a cornerstone of modern Bayesian analysis, has been a major hurdle due to the need to simulate and process massive datasets. Our study tackles these challenges by extending compositional score matching (CSM), a divide-and-conquer strategy for Bayesian updating using diffusion models. We develop a new error-damping estimator to address previous stability issues of CSM when aggregating large numbers of data points. We first verified the numerical stability with up to 100,000 data points on a controlled benchmark. We then evaluated our method on a hierarchical AR model, achieving competitive performance to direct ABI baselines on smaller problem sizes while using less than one full model simulation for larger problem sizes. Finally, we address a large-scale inverse problem in advanced microscopy with over 750,000 parameters, demonstrating its relevance to real scientific applications.

q-bio.QM↗

arXiv2026

BayesFlow 2: Multi-Backend Amortized Bayesian Inference in Python

Modern Bayesian inference involves a mixture of computational methods for estimating, validating, and drawing conclusions from probabilistic models as part of principled workflows. An overarching motif of many Bayesian methods is that they are relatively slow, which often becomes prohibitive when fitting complex models to large data sets. Amortized Bayesian inference (ABI) offers a path to solving the computational challenges of Bayes. ABI trains neural networks on model simulations, rewarding users with rapid inference of any model-implied quantity, such as point estimates, likelihoods, or full posterior distributions. In this work, we present the Python library BayesFlow, Version 2.0, for general-purpose ABI. Along with direct posterior, likelihood, and ratio estimation, the software includes support for multiple popular deep learning backends, a rich collection of generative networks for sampling and density estimation, complete customization and high-level interfaces, as well as new capabilities for hyperparameter optimization, design optimization, and hierarchical modeling. Using a case study on dynamical system parameter estimation, combined with comparisons to similar software, we show that our streamlined, user-friendly workflow has strong potential to support broad adoption.

stat.CO↗

arXiv2026

Robust Amortized Bayesian Inference with Self-Consistency Losses on Unlabeled Data

Amortized Bayesian inference (ABI) with neural networks can solve probabilistic inverse problems orders of magnitude faster than classical methods. However, ABI is not yet sufficiently robust for widespread and safe application. When performing inference on observations outside the scope of the simulated training data, posterior approximations are likely to become highly biased, which cannot be corrected by additional simulations due to the bad pre-asymptotic behavior of current neural posterior estimators. In this paper, we propose a semi-supervised approach that enables training not only on labeled simulated data generated from the model, but also on \textit{unlabeled} data originating from any source, including real data. To achieve this, we leverage Bayesian self-consistency properties that can be transformed into strictly proper losses that do not require knowledge of ground-truth parameters. We test our approach on several real-world case studies, including applications to high-dimensional time-series and image data. Our results show that semi-supervised learning with unlabeled data drastically improves the robustness of ABI in the out-of-simulation regime. Notably, inference remains accurate even when evaluated on observations far away from the labeled and unlabeled data seen during training.

stat.ML↗

arXiv2026

Amortized Bayesian Workflow

Bayesian inference often faces a trade-off between computational speed and sampling accuracy. We propose an adaptive workflow that integrates rapid amortized inference with gold-standard MCMC techniques to achieve a favorable combination of both speed and accuracy when performing inference on many observed datasets. Our approach uses principled diagnostics to guide the choice of inference method for each dataset, moving along the Pareto front from fast amortized sampling via generative neural networks to slower but guaranteed-accurate MCMC when needed. By reusing computations across steps, our workflow synergizes amortized and MCMC-based inference. We demonstrate the effectiveness of this integrated approach on several synthetic and real-world problems with tens of thousands of datasets, showing efficiency gains while maintaining high posterior quality.

cs.LG↗

arXiv2026

Improving the Accuracy of Amortized Model Comparison with Self-Consistency

Amortized Bayesian inference (ABI) offers fast, scalable approximations to posterior densities by training neural surrogates on data simulated from the statistical model. However, ABI methods are highly sensitive to model misspecification: when observed data fall outside the training distribution (generative scope of the statistical models), neural surrogates can behave unpredictably. This makes it a challenge in a model comparison setting, where multiple statistical models are considered, of which at least some are misspecified. Recent work on self-consistency (SC) provides a promising remedy to this issue, accessible even for empirical data (without ground-truth labels). In this work, we investigate how SC can improve amortized model comparison conceptualized in four different ways. Across two synthetic and two real-world case studies, we find that approaches for model comparison that estimate marginal likelihoods through approximate parameter posteriors consistently outperform methods that directly approximate model evidence or posterior model probabilities. SC training improves robustness when the likelihood is available, even under severe model misspecification. The benefits of SC for methods without access of analytic likelihoods are more limited and inconsistent. Our results suggest practical guidance for reliable amortized Bayesian model comparison: prefer parameter posterior-based methods and augment them with SC training on empirical datasets to mitigate extrapolation bias under model misspecification.

stat.ML↗

arXiv2025

Stable Single-Pixel Contrastive Learning for Semantic and Geometric Tasks

We pilot a family of stable contrastive losses for learning pixel-level representations that jointly capture semantic and geometric information. Our approach maps each pixel of an image to an overcomplete descriptor that is both view-invariant and semantically meaningful. It enables precise point-correspondence across images without requiring momentum-based teacher-student training. Two experiments in synthetic 2D and 3D environments demonstrate the properties of our loss and the resulting overcomplete representations.

cs.CV↗

arXiv2025

Testing and Improving the Robustness of Amortized Bayesian Inference for Cognitive Models

Contaminant observations and outliers often cause problems when estimating the parameters of cognitive models, which are statistical models representing cognitive processes. In this study, we test and improve the robustness of parameter estimation using amortized Bayesian inference (ABI) with neural networks. To this end, we conduct systematic analyses on a toy example and analyze both synthetic and real data using a popular cognitive model, the Drift Diffusion Models (DDM). First, we study the sensitivity of ABI to contaminants with tools from robust statistics: the empirical influence function and the breakdown point. Next, we propose a data augmentation or noise injection approach that incorporates a contamination distribution into the data-generating process during training. We examine several candidate distributions and evaluate their performance and cost in terms of accuracy and efficiency loss relative to a standard estimator. Introducing contaminants from a Cauchy distribution during training considerably increases the robustness of the neural density estimator as measured by bounded influence functions and a much higher breakdown point. Overall, the proposed method is straightforward and practical to implement and has a broad applicability in fields where outlier detection or removal is challenging.

stat.ML↗

arXiv2025

Modeling multi-agent motion dynamics in immersive rooms

Immersive rooms are increasingly popular augmented reality systems that support multi-agent interactions within a virtual world. However, despite extensive content creation and technological developments, insights about perceptually-driven social dynamics, such as the complex movement patterns during virtual world navigation, remain largely underexplored. Computational models of motion dynamics can help us understand the underlying mechanism of human interaction in immersive rooms and develop applications that better support spatially distributed interaction. In this work, we propose a new agent-based model of emergent human motion dynamics. The model represents human agents as simple spatial geometries in the room that relocate and reorient themselves based on the salient virtual spatial objects they approach. Agent motion is modeled as an interactive process combining external diffusion-driven influences from the environment with internal self-propelling interactions among agents. Further, we leverage simulation-based inference (SBI) to show that the governing parameters of motion patterns can be estimated from simple observables. Our results indicate that the model successfully captures action-related agent properties but exposes local non-identifiability linked to environmental awareness. We argue that our simulation-based approach paves the way for creating adaptive, responsive immersive rooms -- spaces that adjust their interfaces and interactions based on human collective movement patterns and spatial attention.

cs.HC↗

arXiv2025

Towards Context-Aware Domain Generalization: Understanding the Benefits and Limits of Marginal Transfer Learning

In this work, we analyze the conditions under which information about the context of an input $X$ can improve the predictions of deep learning models in new domains. Following work in marginal transfer learning in Domain Generalization (DG), we formalize the notion of context as a permutation-invariant representation of a set of data points that originate from the same domain as the input itself. We offer a theoretical analysis of the conditions under which this approach can, in principle, yield benefits, and formulate two necessary criteria that can be easily verified in practice. Additionally, we contribute insights into the kind of distribution shifts for which the marginal transfer learning approach promises robustness. Empirical analysis shows that our criteria are effective in discerning both favorable and unfavorable scenarios. Finally, we demonstrate that we can reliably detect scenarios where a model is tasked with unwarranted extrapolation in out-of-distribution (OOD) domains, identifying potential failure cases. Consequently, we showcase a method to select between the most predictive and the most robust model, circumventing the well-known trade-off between predictive performance and robustness.

cs.LG↗

arXiv2025

Simulations in Statistical Workflows

Simulations play important and diverse roles in statistical workflows, for example, in model specification, checking, validation, and even directly in model inference. Over the past decades, the application areas and overall potential of simulations in statistical workflows have expanded significantly, driven by the development of new simulation-based algorithms and exponentially increasing computational resources. In this paper, we examine past and current trends in the field and offer perspectives on how simulations may shape the future of statistical practice.

stat.CO↗

arXiv2025

Bridging Simulations and Observations: New Insights into Galaxy Formation Simulations via Out-of-Distribution Detection and Bayesian Model Comparison

Cosmological simulations are a powerful tool to advance our understanding of galaxy formation and many simulations model key properties of real galaxies. A question that naturally arises for such simulations in light of high-quality observational data is: How close are the models to reality? Due to the high-dimensionality of the problem, many previous studies evaluate galaxy simulations using simplified summary statistics of physical properties. In this work, we combine simulation-based Bayesian model comparison with a novel misspecification detection technique to compare simulated galaxy images of 6 hydrodynamical models observations. Since cosmological simulations are computationally costly, we address the problem of low simulation budgets by first training a $k$-sparse variational autoencoder (VAE) on the abundant dataset of SDSS images. The VAE learns to extract informative latent embeddings and delineates the typical set of real images. To reveal simulation gaps, we then perform out-of-distribution detection (OOD) based on the logits of classifiers trained on the embeddings of simulated images. Finally, we perform amortized Bayesian model comparison using probabilistic classification, identifying the relatively best-performing model along with partial explanations through SHAP values.

astro-ph.GA↗

arXiv2025

Amortized Bayesian Multilevel Models

Multilevel models (MLMs) are a central building block of the Bayesian workflow. They enable joint, interpretable modeling of data across hierarchical levels and provide a fully probabilistic quantification of uncertainty. Despite their well-recognized advantages, MLMs pose significant computational challenges, often rendering their estimation and evaluation intractable within reasonable time constraints. Recent advances in simulation-based inference offer promising solutions for addressing complex probabilistic models using deep generative networks. However, the utility and reliability of deep learning methods for estimating Bayesian MLMs remains largely unexplored, especially when compared with gold-standard samplers. To this end, we explore a family of neural network architectures that leverage the probabilistic factorization of multilevel models to facilitate efficient neural network training and subsequent near-instant posterior inference on unseen datasets. We test our method on several real-world case studies and provide comprehensive comparisons to Stan's gold standard sampler, where possible. Finally, we provide an open-source implementation of our methods to stimulate further research in the nascent field of amortized Bayesian inference.

stat.ML↗