LADB: Latent Aligned Diffusion Bridges for Semi-Supervised Domain Translation

2025-09-10

AI Summary

Diffusion models face high retraining costs and heavy reliance on large-scale paired data in data-scarce scenarios. To address this, we propose LADB, a semi-supervised domain translation framework that aligns source and target domains within a shared latent space using only a small number of paired samples. Its core innovation is the Latent Alignment Diffusion Bridge (LADB), which establishes a deterministic cross-domain mapping by coupling latent variables from pre-trained diffusion models—without requiring full supervision. The method integrates class-conditional style transfer with collaborative latent optimization, enabling natural extension to multi-source and multi-target settings. Evaluated on depth-to-image translation, LADB significantly improves generation fidelity and diversity. Experiments demonstrate its effectiveness, generality, and scalability in settings where annotations are costly and data is incomplete or partially observed.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Web Mining and Content Analysis: Large pretrained models with web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Diffusion models excel at generating high-quality outputs but face challenges in data-scarce domains, where exhaustive retraining or costly paired data are often required. To address these limitations, we propose Latent Aligned Diffusion Bridges (LADB), a semi-supervised framework for sample-to-sample translation that effectively bridges domain gaps using partially paired data. By aligning source and target distributions within a shared latent space, LADB seamlessly integrates pretrained source-domain diffusion models with a target-domain Latent Aligned Diffusion Model (LADM), trained on partially paired latent representations. This approach enables deterministic domain mapping without the need for full supervision. Compared to unpaired methods, which often lack controllability, and fully paired approaches that require large, domain-specific datasets, LADB strikes a balance between fidelity and diversity by leveraging a mixture of paired and unpaired latent-target couplings. Our experimental results demonstrate superior performance in depth-to-image translation under partial supervision. Furthermore, we extend LADB to handle multi-source translation (from depth maps and segmentation masks) and multi-target translation in a class-conditioned style transfer task, showcasing its versatility in handling diverse and heterogeneous use cases. Ultimately, we present LADB as a scalable and versatile solution for real-world domain translation, particularly in scenarios where data annotation is costly or incomplete.

Problem

• Semi-supervised translation with limited paired data
• Bridging domain gaps using partial supervision
• Enabling controllable cross-domain mapping without full retraining

Innovation

• Semi-supervised framework using partially paired data
• Aligns source and target in shared latent space
• Integrates pretrained diffusion with latent aligned model