A multifidelity approximate Bayesian computation with pre-filtering

2026-02-02

AI Summary

This work addresses the high computational cost of Approximate Bayesian Computation (ABC), which stems from its reliance on extensive simulations and the lack of efficient sampling strategies in multi-fidelity modeling. The authors propose the first multi-fidelity ABC framework incorporating a pre-screening mechanism, which integrates stratified importance sampling with an adaptive sequential Monte Carlo algorithm to substantially reduce computational overhead. A theoretical analysis establishes quantitative relationships among posterior concentration, error upper bounds, and the pre-screening criterion, along with a strategy for assessing model applicability. Numerical experiments demonstrate that the proposed method achieves significant gains in computational efficiency while preserving inference accuracy. To facilitate community adoption, the authors release an open-source R package, MAPS.

Technology Category

Search and Optimization: Sampling/Simulation-based SearchReasoning under Uncertainty: Relational Probabilistic ModelsMachine Learning: Calibration & Uncertainty Quantification

Application Category

Search and Retrieval-Augmented AI: Web evaluation methodologies and metricsGraph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphsUser Modeling, Personalization and Recommendation: Metrics for user behavior and evaluating success

Abstract

Approximate Bayesian Computation (ABC) methods often require extensive simulations, resulting in high computational costs. This paper focuses on multifidelity simulation models and proposes a pre-filtering hierarchical importance sampling algorithm. Under mild assumptions, we theoretically prove that the proposed algorithm satisfies posterior concentration properties, characterize the error upper bound and the relationship between algorithmic efficiency and pre-filtering criteria. Additionally, we provide a practical strategy to assess the suitability of multifidelity models for the proposed method. Finally, we develop a multifidelity ABC sequential Monte Carlo with adaptive pre-filtering strategy. Numerical experiments are used to demonstrate the effectiveness of the proposed approach. We develop an R package that is available at https://github.com/caofff/MAPS

Problem

• Approximate Bayesian Computation
• multifidelity simulation
• computational cost
• pre-filtering
• Bayesian inference

Innovation

• multifidelity
• approximate Bayesian computation
• pre-filtering
• sequential Monte Carlo
• importance sampling