---
title: "DSSL-Hash: Dynamic Semantic Structure Learning for Unsupervised Cross-Modal Hashing"
authors:
  - "Fan Yang"
  - "Tongxuan Pei"
  - "Yuanzhi Zhao"
  - "Yudong Zhao"
doi: "10.24963/ijcai.2026/219"
url: "https://ijcai.org/proceedings/2026/219"
pdf: "https://ijcai.org/proceedings/2026/0219.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "1965-1973"
track: "Main Track"
keywords:
  - "Machine Learning: Multi-modal learning"
  - "Natural Language Processing: Information retrieval and text mining"
  - "Computer Vision: Image and video retrieval"
  - "Data Mining: Information retrieval"
  - "Machine Learning: Multi-view learning"
---

# DSSL-Hash: Dynamic Semantic Structure Learning for Unsupervised Cross-Modal Hashing

Fan Yang, Tongxuan Pei, Yuanzhi Zhao, Yudong Zhao

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 1965-1973. DOI: [10.24963/ijcai.2026/219](https://doi.org/10.24963/ijcai.2026/219)

## Abstract

A central bottleneck in unsupervised cross-modal hashing is that both similarity guidance and objective weighting are typically static, while the semantic structure induced during training is inherently dynamic. We present DSSL-Hash, which turns semantic structure learning into an evolving process rather than a one-shot preprocessing step. Specifically, DSSL-Hash distills semantic cues from a vision–language foundation model  and constructs a dynamic perception graph via an adaptive similarity matrix, enabling graph propagation to capture higher-order relations beyond pairwise matching. We further reshape Hamming-space learning with a semantic-channel constraint that explicitly regulates distances for partially similar pairs, reducing semantic distortion after binarization. Finally, we develop a Retrieval-Aware Adaptive Scheduler (RAAS) that leverages retrieval feedback to co-adjust modality interactions and objective weights, achieving robust optimization without extensive manual hyper-parameter search. Extensive experiments on MS COCO, NUS-WIDE, and MIRFLICKR-25K across multiple code lengths demonstrate that DSSL-Hash consistently outperforms recent state-of-the-art unsupervised cross-modal hashing methods.