Latent Discriminative Space with Inline Constraints Cross-Modal Hashing
Cross-modal hashing utilize the advantages of hash codes to enable flexible retrieval across different modalities, greatly improving the retrieval efficiency of heterogeneous modes. Thereinto, most existing approach do not fully take modal intrinsic semantic properties and semantic category structure into consideration during learning the latent subspace. To alleviate these problems, in this paper, we propose a novel cross-modal hashing approach, i.e., Latent Discriminative Space with Inline Constraints cross-modal hashing, LDSIC for short. We integrate the latent subspaces for each individual modality into the hash codes learning process which aims to learn more discriminative hash codes and preserve the semantic similarity. LDSIC approach utilizes label space as the discriminative space for improved retrieval performance, and each modality data is decomposed individually to explore latent semantic spaces, through the label subspace and mathematical constraint, the specific representations of each modality data are connected to make similar data from different modalities closer. In this way, hash codes with discriminability and stability are learned, moreover, an effective iterative alternative optimization scheme is developed to solve the NP-hard optimization problem. Through experimental on the three benchmark datasets demonstrate that our approach can achieve significantly retrieval precision out-performs several state-of-the-art approaches
Year of publication: |
[2022]
|
---|---|
Authors: | Yang, Fan ; Ding, Xiao-Jian ; Liu, Yu-Feng ; Ma, Fuming ; Cao, Jie |
Publisher: |
[S.l.] : SSRN |
Saved in:
freely available
Saved in favorites
Similar items by person
-
Valley polarized insulator-metal transition and valley filtering effect in graphene
Cao, Jie, (2015)
-
A quantitative assessment of energy strategy evolution in China and US
Zhang, Huiming, (2011)
-
ON ADDITIVE CONSISTENT PROPERTIES OF THE INTUITIONISTIC FUZZY PREFERENCE RELATION
GONG, ZAIWU, (2010)
- More ...