Image Super-Resolution with Guarantees via Conformal Generative Models

2025-02-12

AI Summary

This work addresses the lack of reliable and interpretable uncertainty quantification in generative image super-resolution models. We propose the first conformal prediction framework for this task—Confidence Masking—a black-box method that requires no access to model internals, relying solely on API outputs and a customizable local similarity metric to produce pixel-wise, theoretically guaranteed uncertainty estimates. Leveraging nonparametric calibration and carefully designed post-processing, our approach ensures robustness against data leakage and metric portability across diverse similarity measures. Evaluated on multiple benchmarks, it strictly controls local fidelity errors (e.g., deviations in PSNR and SSIM) while significantly enhancing result credibility and interpretability. Our framework establishes a novel paradigm for trustworthy deployment of generative models in safety-critical applications.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Adversarial Attacks & Robustness
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

The increasing use of generative ML foundation models for image super-resolution calls for robust and interpretable uncertainty quantification methods. We address this need by presenting a novel approach based on conformal prediction techniques to create a"confidence mask"capable of reliably and intuitively communicating where the generated image can be trusted. Our method is adaptable to any black-box generative model, including those locked behind an opaque API, requires only easily attainable data for calibration, and is highly customizable via the choice of a local image similarity metric. We prove strong theoretical guarantees for our method that span fidelity error control (according to our local image similarity metric), reconstruction quality, and robustness in the face of data leakage. Finally, we empirically evaluate these results and establish our method's solid performance.

Problem

• Uncertainty quantification in image super-resolution
• Confidence mask for generated images
• Adaptable to any black-box generative model

Innovation

• Conformal prediction for uncertainty
• Adaptable to black-box models
• Local image similarity customization