Flow-Based Density Ratio Estimation for Intractable Distributions with Applications in Genomics

2026-02-27

AI Summary

This work addresses the computational challenge of estimating density ratios between intractable distributions by proposing a conditional-aware flow matching framework. By directly modeling the dynamics of density ratios along generative trajectories, the method circumvents the need for costly likelihood integrations over individual distributions. It introduces flow matching—a technique previously unexplored in this context—into density ratio estimation, leveraging a single ordinary differential equation (ODE)-based generative model to jointly characterize density ratios across multiple conditional distributions. This unified approach substantially enhances both computational efficiency and modeling flexibility. The framework achieves high-accuracy, closed-form density ratio estimates on synthetic data and demonstrates practical utility in single-cell genomics, where it is successfully applied to treatment effect estimation and batch correction.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Sampling/Simulation-based Search
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Estimating density ratios between pairs of intractable data distributions is a core problem in probabilistic modeling, enabling principled comparisons of sample likelihoods under different data-generating processes across conditions and covariates. While exact-likelihood models such as normalizing flows offer a promising approach to density ratio estimation, naive flow-based evaluations are computationally expensive, as they require simulating costly likelihood integrals for each distribution separately. In this work, we leverage condition-aware flow matching to derive a single dynamical formulation for tracking density ratios along generative trajectories. We demonstrate competitive performance on simulated benchmarks for closed-form ratio estimation, and show that our method supports versatile tasks in single-cell genomics data analysis, where likelihood-based comparisons of cellular states across experimental conditions enable treatment effect estimation and batch correction evaluation.

Problem

• density ratio estimation
• intractable distributions
• normalizing flows
• single-cell genomics
• likelihood comparison

Innovation

• flow-based density ratio estimation
• intractable distributions
• condition-aware flow matching
• single-cell genomics
• likelihood-free inference