A novel two-stage parameter estimation framework integrating Approximate Bayesian Computation and Machine Learning: The ABC-RF-rejection algorithm

2025-07-02

AI Summary

Parameter estimation for complex dynamic models—such as stochastic or non-analytic systems—is often computationally prohibitive due to expensive simulations. To address this, we propose a two-stage ABC-RF-rejection framework: first, a random forest classifier pre-filters simulated datasets to identify promising regions of the parameter space; second, Approximate Bayesian Computation (ABC) rejection sampling is applied only to this refined subset. This approach innovatively integrates supervised learning into the ABC pipeline, substantially improving the efficiency of likelihood-free inference in high-dimensional parameter spaces. Experiments on deterministic SIR and spatially explicit stochastic epidemic models demonstrate estimation accuracy comparable to standard ABC, yet with an order-of-magnitude reduction in computational cost. Furthermore, the method successfully estimates spatial transmission parameters for cassava brown streak disease in Uganda, confirming its practical utility and scalability to real-world epidemiological applications.

Technology Category

• Reasoning under Uncertainty: Stochastic Optimization
• Machine Learning: Probabilistic Circuits and Graphical Models
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

We introduce a novel two-stage parameter estimation framework designed to improve computational efficiency in settings involving complex, stochastic, or analytically intractable dynamic models. The proposed method, termed extit{ABC-RF-rejection}, integrates Approximate Bayesian Computation (ABC) rejection sampling with Random Forest (RF) classification to efficiently screen parameter sets that produce simulations consistent with observed data. We evaluate the performance of this approach using both a deterministic Susceptible-Infected-Removed (SIR) epidemic model and a spatially explicit stochastic epidemic model. Results indicate that ABC-RF-rejection achieves substantial gains in computational efficiency while maintaining parameter inference accuracy comparable with standard ABC rejection methods. Finally, we apply the algorithm to estimate parameters governing the spatial spread of cassava brown streak disease (CBSD) in Nakasongola district, Uganda.

Problem

• Improving computational efficiency in complex dynamic models
• Integrating ABC and ML for parameter estimation
• Estimating disease spread parameters accurately and efficiently

Innovation

• Two-stage parameter estimation framework
• Integrates ABC and Random Forest
• Improves computational efficiency significantly