Simulation-Based Fitting of Intractable Models via Sequential Sampling and Local Smoothing

TL;DR

This paper introduces a versatile simulation-based algorithm for fitting complex generative models with intractable likelihoods, combining global and local search techniques to efficiently estimate parameters with minimal prior information.

Contribution

The paper proposes a novel, comprehensive algorithm that integrates global search and local smoothing for fitting intractable models, with an available R package implementation.

Findings

Demonstrates strong performance compared to alternative methods

Effective in identifying solution regions with limited prior info

Provides a practical tool for complex model fitting

Abstract

This paper presents a comprehensive algorithm for fitting generative models whose likelihood, moments, and other quantities typically used for inference are not analytically or numerically tractable. The proposed method aims to provide a general solution that requires only limited prior information on the model parameters. The algorithm combines a global search phase, aimed at identifying the region of the solution, with a local search phase that mimics a trust region version of the Fisher scoring algorithm for computing a quasi-likelihood estimator. Comparisons with alternative methods demonstrate the strong performance of the proposed approach. An R package implementing the algorithm is available on CRAN.