Geometry Preserving Loss Functions Promote Improved Adaptation of Blackbox Generative Model

2026-04-26

AI Summary

This work addresses the challenge of adapting black-box generative models—whose weights are inaccessible and fine-tuning is prohibitively expensive—to specific target domains. The authors propose an end-to-end domain adaptation framework that redefines the role of GAN inversion by incorporating a geometry-preserving loss in the StyleGAN latent space. This loss explicitly constrains pairwise sample distances within the tangent space, enabling the learning of precise latent representations that better align with the target distribution. By reformulating GAN inversion as a domain adaptation mechanism, the method significantly outperforms conventional loss functions and effectively enhances the generation quality and adaptability of black-box models under realistic distribution shifts.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Natural Language Processing: Generation

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Adaptation of blackbox generative models has been widely studied recently through the exploration of several methods including generator fine-tuning, latent space searches, leveraging singular value decomposition, and so on. However, adapting large-scale generative AI tools to specific use cases continues to be challenging, as many of these industry-grade models are not made widely available. The traditional approach of fine-tuning certain layers of a generative network is not feasible due to the expense of storing and fine-tuning generative models, as well as the restricted access to weights and gradients. Recognizing these challenges, we propose a novel end-to-end pipeline aimed at domain adaptation by leveraging geometry-preserving loss functions in conjunction to pre-trained generative adversarial networks (GANs). Our method rethinks the problem of adaptation by re-contextualizing the role of GAN inversion in obtaining accurate latent space representations. Extending the ability of existing state-of-the-art inverters, we preserve pair-wise distances between tangent spaces to successfully train a latent generative model to produce samples from the target distribution. We evaluate our proposed pipeline on StyleGANs with real distribution shifts and demonstrate that the introduction of the geometry preserving loss function lends to improved adaptation of generative models compared to other traditional loss functions.

Problem

• blackbox generative model
• domain adaptation
• geometry preserving
• GAN inversion
• latent space

Innovation

• geometry-preserving loss
• blackbox generative model adaptation
• GAN inversion
• latent space geometry
• domain adaptation