Fast and Robust Likelihood-Guided Diffusion Posterior Sampling with Amortized Variational Inference

2026-02-06

AI Summary

This work addresses the high computational cost of zero-shot posterior sampling in diffusion models for inverse problems and the limited generalization of existing amortized methods to unseen degradation operators. The authors propose a variational inference–based amortization strategy that explicitly models likelihood guidance and amortizes the inner optimization loop, thereby significantly accelerating inference while preserving the flexibility of zero-shot approaches. By uniquely integrating amortized inference with explicit likelihood guidance, the method achieves efficient reconstruction on in-distribution degradations and demonstrates remarkable robustness and reconstruction quality under out-of-distribution or previously unseen degradation operators, consistently outperforming both current amortized and zero-shot baselines.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Probabilistic Inference

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Zero-shot diffusion posterior sampling offers a flexible framework for inverse problems by accommodating arbitrary degradation operators at test time, but incurs high computational cost due to repeated likelihood-guided updates. In contrast, previous amortized diffusion approaches enable fast inference by replacing likelihood-based sampling with implicit inference models, but at the expense of robustness to unseen degradations. We introduce an amortization strategy for diffusion posterior sampling that preserves explicit likelihood guidance by amortizing the inner optimization problems arising in variational diffusion posterior sampling. This accelerates inference for in-distribution degradations while maintaining robustness to previously unseen operators, thereby improving the trade-off between efficiency and flexibility in diffusion-based inverse problems.

Problem

• diffusion posterior sampling
• inverse problems
• amortized inference
• likelihood guidance
• robustness

Innovation

• amortized variational inference
• likelihood-guided diffusion
• posterior sampling
• inverse problems
• zero-shot generalization