arXiv:2602.09161v2 Announce Type: replace 
Abstract: Simulation-based inference (SBI) enables amortized Bayesian inference by first training a neural posterior estimator (NPE) on prior-simulator pairs, typically through low-dimensional summary statistics, which can then be cheaply reused for fast inference by querying it on new test observations. Because NPE is estimated under the training data distribution, it is susceptible to misspecification when observations deviate from the training distribution. Many robust SBI approaches address this by modifying NPE training or introducing error models, coupling robustness to the inference network and compromising amortization and modularity. We introduce minimum-distance summaries, a plug-in robust NPE method that adapts queried test-time summaries independently of the pretrained NPE. Leveraging the maximum mean discrepancy (MMD) as a distance between observed data and a summary-conditional predictive distribution, the adapted summary inherits strong robustness properties from the MMD. We demonstrate that the algorithm can be implemented efficiently with random Fourier feature approximations, yielding a lightweight, model-free test-time adaptation procedure. We provide theoretical guarantees for the robustness of our algorithm and empirically evaluate it on a range of synthetic and real-world tasks, demonstrating substantial robustness gains with minimal additional overhead.

دراسة حديثة تقدم ملخصات المسافة الدنيا كطريقة قوية لتقدير ما بعد الشبكة العصبية، حيث تتناول القيود المفروضة على مقدرات ما بعد الشبكة العصبية التقليدية (NPE) التي تكون حساسة للاختلافات عن توزيعات بيانات التدريب. تستخدم هذه الطريقة التباين المتوسط الأقصى (MMD) لتعديل الملخصات في وقت الاختبار بشكل مستقل، مما يعزز موثوقية الاستدلال القائم على المحاكاة (SBI).

Un estudio reciente introduce resúmenes de distancia mínima como un método robusto para la estimación posterior neuronal, abordando las limitaciones de los estimadores posteriores neuronales (NPE) tradicionales que son sensibles a las desviaciones de las distribuciones de datos de entrenamiento. Este método utiliza la discrepancia media máxima (MMD) para modificar de manera adaptativa los resúmenes en el tiempo de prueba, mejorando así la fiabilidad de la inferencia basada en simulaciones (SBI).

Une étude récente présente des résumés de distance minimale comme une méthode robuste pour l'estimation postérieure neuronale, abordant les limitations des estimateurs postérieurs neuronaux (NPE) traditionnels qui sont sensibles aux écarts par rapport aux distributions de données d'entraînement. Cette méthode utilise la distance moyenne maximale (MMD) pour modifier de manière adaptative les résumés en temps de test, améliorant ainsi la fiabilité de l'inférence basée sur la simulation (SBI).

A recent study introduces minimum-distance summaries as a robust method for neural posterior estimation, addressing the limitations of traditional neural posterior estimators (NPE) that are sensitive to deviations from training data distributions. This method utilizes maximum mean discrepancy (MMD) to adaptively modify test-time summaries, enhancing the reliability of simulation-based inference (SBI).