arXiv:2512.10445v1 Announce Type: new 
Abstract: We consider a regression setting where observations are collected in different environments modeled by different data distributions. The field of out-of-distribution (OOD) generalization aims to design methods that generalize better to test environments whose distributions differ from those observed during training. One line of such works has proposed to minimize the maximum risk across environments, a principle that we refer to as MaxRM (Maximum Risk Minimization). In this work, we introduce variants of random forests based on the principle of MaxRM. We provide computationally efficient algorithms and prove statistical consistency for our primary method. Our proposed method can be used with each of the following three risks: the mean squared error, the negative reward (which relates to the explained variance), and the regret (which quantifies the excess risk relative to the best predictor). For MaxRM with regret as the risk, we prove a novel out-of-sample guarantee over unseen test distributions. Finally, we evaluate the proposed methods on both simulated and real-world data.

قدمت دراسة حديثة متغيرات من الغابات العشوائية استنادًا إلى مبدأ الحد الأقصى لتقليل المخاطر (MaxRM)، الذي يهدف إلى تحسين أداء الانحدار عبر توزيعات بيانات مختلفة. تتناول هذه الطريقة تحدي التعميم خارج التوزيع، حيث توفر خوارزميات فعالة من حيث الحساب وتثبت التناسق الإحصائي للطرق المقترحة.

Un estudio reciente ha introducido variantes de bosques aleatorios basadas en el principio de Minimización del Riesgo Máximo (MaxRM), que busca mejorar el rendimiento de la regresión a través de diferentes distribuciones de datos. Este enfoque aborda el desafío de la generalización fuera de distribución, proporcionando algoritmos computacionalmente eficientes y demostrando la consistencia estadística de los métodos propuestos.

Une étude récente a introduit des variantes des forêts aléatoires basées sur le principe de la minimisation du risque maximum (MaxRM), visant à améliorer la performance de régression à travers différentes distributions de données. Cette approche répond au défi de la généralisation hors distribution, en fournissant des algorithmes efficaces sur le plan computationnel et en prouvant la cohérence statistique des méthodes proposées.

A recent study has introduced variants of random forests based on the principle of Maximum Risk Minimization (MaxRM), which aims to enhance regression performance across different data distributions. This approach addresses the challenge of out-of-distribution generalization, providing computationally efficient algorithms and proving statistical consistency for the proposed methods.

Maximum Risk Minimization with Random Forests

Was this article worth reading? Share it

LucidQuery AI

Hypertune

Portfolio Backtest

MatosShere

MicroEstimates

Offramp

Ready to build your own newsroom?