Cross-Domain Diffusion with Progressive Alignment for Efficient Adaptive Retrieval

TL;DR

COUPLE addresses unsupervised domain-adaptive hashing by modeling cross-domain transfer as a graph diffusion process to identify low-noise target samples and guide robust discriminative hash learning. It further introduces progressive domain alignment via hierarchical Mixup along cross-domain random-walk paths, enabling gradual, multi-level domain adaptation. The approach yields state-of-the-art cross-domain retrieval performance across multiple benchmarks, with improved robustness to target-domain noise, faster hash-based retrieval, and strong backbone generalization. These results demonstrate the practical impact of diffusion-guided sample selection and progressive alignment for scalable, noise-resilient cross-domain retrieval.

Abstract

Unsupervised efficient domain adaptive retrieval aims to transfer knowledge from a labeled source domain to an unlabeled target domain, while maintaining low storage cost and high retrieval efficiency. However, existing methods typically fail to address potential noise in the target domain, and directly align high-level features across domains, thus resulting in suboptimal retrieval performance. To address these challenges, we propose a novel Cross-Domain Diffusion with Progressive Alignment method (COUPLE). This approach revisits unsupervised efficient domain adaptive retrieval from a graph diffusion perspective, simulating cross-domain adaptation dynamics to achieve a stable target domain adaptation process. First, we construct a cross-domain relationship graph and leverage noise-robust graph flow diffusion to simulate the transfer dynamics from the source domain to the target domain, identifying lower noise clusters. We then leverage the graph diffusion results for discriminative hash code learning, effectively learning from the target domain while reducing the negative impact of noise. Furthermore, we employ a hierarchical Mixup operation for progressive domain alignment, which is performed along the cross-domain random walk paths. Utilizing target domain discriminative hash learning and progressive domain alignment, COUPLE enables effective domain adaptive hash learning. Extensive experiments demonstrate COUPLE's effectiveness on competitive benchmarks.

Figures (11)

• Figure 1: Overview of COUPLE. Our objective is to learn domain-adaptive hash codes for the target domain. Specifically, we leverage cross-domain diffusion to simulate the transfer dynamics and explore the target domain robustly. Furthermore, we perform hierarchical Mixup learning along cross-domain random walk paths to achieve progressive domain alignment.
• Figure 2: Hierarchical Mixup. Low-level Mixup is for cross-domain pixel-level alignment, and high-level Mixup is to capture cross-domain semantic information. They contribute to the comprehensive adaptive hash codes under the different level domain shifts.
• Figure 3: Precision-recall curves with $64$ bits hash code on Office-Home and Office-31 datasets.
• Figure 4: Top-N precision curves with $64$ bits hash code on Office-Home and Office31 datasets.
• Figure 5: Top-N recall curves with $64$ bits hash code on Office-Home and Office31 datasets.

Theorems & Definitions (1)

• Theorem 3.1