Pith. sign in

REVIEW 2 minor 52 references

APKH uses prompt-optimized attribute kernels from vision-language models to align modalities in a Hamming space for data-efficient retrieval on unseen categories.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.3

2026-07-02 07:02 UTC pith:3UMVDBUF

load-bearing objection APKH tries to fix data scarcity in unsupervised cross-modal hashing by pulling attribute priors from VLMs and smoothing contrastive alignment with kernels, but the gains look incremental and rest on unverified experimental details.

arxiv 2607.00379 v1 pith:3UMVDBUF submitted 2026-07-01 cs.IR cs.CV

Attribute-Prompted Kernel Hashing for Unsupervised Data-Efficient Cross-Modal Retrieval

Runhao Li , Xiaoxu Ma , Zhenyu Weng , Yue Zhang , Guibo Luo , Huiping Zhuang , Zhiping Lin , Yap-Peng Tan This is my paper
classification cs.IR cs.CV
keywords unsupervised cross-modal hashingdata-efficient retrievalattribute promptingkernel mappingcontrastive alignmentvision-language modelsgeneralization to unseen categories
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes Attribute-Prompted Kernel Hashing (APKH) to perform unsupervised cross-modal retrieval when aligned image-text pairs are scarce. It leverages generalized attribute priors from vision-language foundation models and optimizes them through prompt learning. The CAKM module maps data via hyperspherical RBF kernels to capture invariant semantics, while KSCA models sparse pairs as kernel distributions to smooth the alignment and reduce overfitting. This enables better generalization from seen training data to unseen categories in constrained scenarios.

Core claim

APKH constructs a compact, modality-aligned Hamming space driven by the generalized attribute priors of vision-language foundation models. CAKM formulates alignment through hyperspherical Radial Basis Function kernel mapping and optimizes dynamic attribute kernels via prompt learning. KSCA extends point-to-point contrastive learning by treating limited paired data as continuous kernel distributions, which smooths the modality gap and alleviates overfitting to sparse correlations.

What carries the argument

Context-optimized Attribute Kernel Mapping (CAKM) combined with Kernel-Smoothed Contrastive Alignment (KSCA), which together optimize prompt-driven attribute kernels on hyperspherical RBF mappings and model pairs as distributions to produce modality-invariant semantics.

Load-bearing premise

Generalized attribute priors from vision-language models can be prompt-optimized to yield modality-invariant semantics that remain effective when the number of aligned training pairs is severely limited.

What would settle it

Apply APKH and prior hashing methods to a cross-modal dataset restricted to a small number of aligned pairs, then measure retrieval accuracy on unseen categories; if APKH shows no consistent gain over baselines, the data-efficiency and generalization claims fail.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

0 major / 2 minor

Summary. The manuscript proposes Attribute-Prompted Kernel Hashing (APKH) for unsupervised data-efficient cross-modal retrieval. It constructs a modality-aligned Hamming space using generalized attribute priors from vision-language foundation models. The core contributions are Context-optimized Attribute Kernel Mapping (CAKM), which performs hyperspherical RBF kernel mapping with prompt-optimized dynamic attribute kernels, and Kernel-Smoothed Contrastive Alignment (KSCA), which models limited paired data as continuous kernel distributions to reduce overfitting to sparse correlations. Experiments are claimed to show outperformance over SOTA hashing methods on seen-to-unseen category retrieval under data-constrained conditions.

Significance. If the empirical claims hold, the work addresses a practically relevant gap in cross-modal hashing where large-scale aligned pairs are unavailable due to cost or privacy constraints. The kernel-distribution smoothing of contrastive alignment offers a plausible mechanism for improving generalization beyond pointwise losses. No machine-checked proofs or parameter-free derivations are present, but the approach is internally consistent with its stated goals.

minor comments (2)
  1. Abstract: the phrase 'generalized attribute priors' is used without a brief definition or reference to how they are extracted from the foundation model; a short parenthetical would improve readability for readers outside the immediate sub-area.
  2. The manuscript should clarify in the experimental section whether the prompt optimization in CAKM is performed jointly with hashing or in a separate stage, as this affects reproducibility.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive evaluation of our manuscript and the recommendation for minor revision. The assessment that the work addresses a relevant gap in data-efficient cross-modal hashing is appreciated. No major comments were raised in the report.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper describes a technical approach consisting of CAKM (hyperspherical RBF kernel mapping with prompt-optimized attribute kernels) and KSCA (kernel-smoothed contrastive alignment) to achieve modality-invariant semantics for data-efficient cross-modal hashing. No equations, fitted parameters, or derivation steps are visible in the provided text that reduce a claimed prediction or result to an input by construction. No self-citation chains, uniqueness theorems, or ansatzes are invoked as load-bearing elements. The central claims rest on empirical performance in seen-to-unseen retrieval under data constraints, which is an external validation question rather than an internal definitional loop. The derivation chain is therefore self-contained against the stated goals.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only; no explicit free parameters, axioms, or invented entities are stated. The method implicitly assumes that prompt learning on foundation-model attributes yields modality-invariant kernels without additional supervision.

pith-pipeline@v0.9.1-grok · 5792 in / 1160 out tokens · 27326 ms · 2026-07-02T07:02:07.614175+00:00 · methodology

0 comments
read the original abstract

Unsupervised cross-modal hashing enables efficient retrieval of semantically related instances across different modalities without requiring manual semantic annotation. However, existing unsupervised methods rely heavily on large-scale image-text pairs. Collecting such data can be costly, particularly in scenarios where well-aligned pairs are scarce due to privacy and specialized constraints. More critically, existing methods tend to overfit to seen training data, restricting their generalization performance on unseen categories that the constrained training data cannot cover. To address these limitations, we propose Attribute-Prompted Kernel Hashing (APKH), a novel data-efficient approach that constructs a compact, modality-aligned Hamming space driven by the generalized attribute priors of vision-language foundation models. Specifically, APKH introduces two core modules: Context-optimized Attribute Kernel Mapping (CAKM) and Kernel-Smoothed Contrastive Alignment (KSCA). CAKM formulates cross-modal alignment through hyperspherical Radial Basis Function kernel mapping, optimizing dynamic attribute kernels via prompt learning to capture modality-invariant semantics. Furthermore, KSCA extends conventional point-to-point contrastive learning by modeling limited paired data as continuous kernel distributions. This explicit smoothing of the modality gap alleviates overfitting to sparse pairwise correlations. Extensive experiments demonstrate that APKH outperforms state-of-the-art hashing methods in the challenging cross-modal retrieval tasks from seen to unseen categories under data-constrained scenarios.

Figures

Figures reproduced from arXiv: 2607.00379 by Guibo Luo, Huiping Zhuang, Runhao Li, Xiaoxu Ma, Yap-Peng Tan, Yue Zhang, Zhenyu Weng, Zhiping Lin.

Figure 1
Figure 1. Figure 1: The inference pipelines of existing methods vs. our proposed approach. While existing methods (left) map inputs using separate modality-specific hash [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Framework of our proposed method. Heterogeneous inputs (images and texts) are first encoded by frozen CLIP models. Our method introduces [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Retrieval performance (mAP) of the proposed APKH and state-of [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Precision-Recall (PR) curves for image-to-text (I2T) and text-to-image (T2I) retrieval tasks on NUS-WIDE and Wikipedia datasets. The results compare [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The t-SNE visualizations of the relaxed hash codes on Pascal Sentence and Wikipedia datasets. The plots compare the feature distributions of CAGAN, [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Impact of the number of attributes (K) on cross-modal retrieval performance. The results are averaged across all datasets. 1 2 4 8 16 L 0.425 0.450 0.475 0.500 0.525 0.550 0.575 0.600 mAP Performance vs Number of Context Tokens (L) Seen Unseen Average [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Impact of the number of context tokens ( [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

52 extracted references · 52 canonical work pages · 3 internal anchors

  1. [1]

    Deep cross-modal hashing,

    Q.-Y . Jiang and W.-J. Li, “Deep cross-modal hashing,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 3232–3240

    work page 2017

  2. [2]

    Ckdh: Clip- based knowledge distillation hashing for cross-modal retrieval,

    J. Li, W. K. Wong, L. Jiang, X. Fang, S. Xie, and Y . Xu, “Ckdh: Clip- based knowledge distillation hashing for cross-modal retrieval,”IEEE Transactions on Circuits and Systems for Video Technology, 2024

    work page 2024

  3. [3]

    Neighborhood learning from noisy labels for cross-modal retrieval,

    R. Li, Z. Weng, H. Zhuang, Y . Chen, and Z. Lin, “Neighborhood learning from noisy labels for cross-modal retrieval,” in2023 IEEE International Symposium on Circuits and Systems (ISCAS). IEEE, 2023, pp. 1–5

    work page 2023

  4. [4]

    Deep neighborhood-preserving hashing with quadratic spherical mutual infor- mation for cross-modal retrieval,

    Q. Qin, Y . Huo, L. Huang, J. Dai, H. Zhang, and W. Zhang, “Deep neighborhood-preserving hashing with quadratic spherical mutual infor- mation for cross-modal retrieval,”IEEE Transactions on Multimedia, vol. 26, pp. 6361–6374, 2024

    work page 2024

  5. [5]

    Dual prototypes-based personalized federated adversarial cross-modal hashing,

    L. Gu, X. Shen, J. Sun, Y . Liu, J. Li, Z. Li, S.-C. S. Cheung, and W. Wan, “Dual prototypes-based personalized federated adversarial cross-modal hashing,”IEEE Transactions on Circuits and Systems for Video Technology, 2025

    work page 2025

  6. [6]

    Pfedlah: Personalized feder- ated learning with lookahead for adaptive cross-modal hashing,

    Y . Chen, H. Lin, Z. Yang, and J. Long, “Pfedlah: Personalized feder- ated learning with lookahead for adaptive cross-modal hashing,”IEEE Transactions on Circuits and Systems for Video Technology, 2025

    work page 2025

  7. [7]

    Cross-modal hashing via diverse instances matching,

    J. Tu, X. Liu, Z. Huang, Y . Hao, R. Hong, and M. Wang, “Cross-modal hashing via diverse instances matching,”IEEE Transactions on Image Processing, 2025

    work page 2025

  8. [8]

    Pic-cmh: Efficient prompt-infused continual cross-modal hashing,

    F. Li, W. Liu, T. Wang, L. Zhu, and X. Chang, “Pic-cmh: Efficient prompt-infused continual cross-modal hashing,”IEEE Transactions on Multimedia, vol. 28, pp. 2654–2667, 2026

    work page 2026

  9. [9]

    Learning hash functions for cross-view similarity search,

    S. Kumar and R. Udupa, “Learning hash functions for cross-view similarity search,” inTwenty-second international joint conference on artificial intelligence, 2011

    work page 2011

  10. [10]

    Joint-modal distribution- based similarity hashing for large-scale unsupervised deep cross-modal retrieval,

    S. Liu, S. Qian, Y . Guan, J. Zhan, and L. Ying, “Joint-modal distribution- based similarity hashing for large-scale unsupervised deep cross-modal retrieval,” inProceedings of the 43rd international ACM SIGIR confer- ence on research and development in information retrieval, 2020, pp. 1379–1388

    work page 2020

  11. [11]

    Deep joint-semantics reconstructing hashing for large-scale unsupervised cross-modal retrieval,

    S. Su, Z. Zhong, and C. Zhang, “Deep joint-semantics reconstructing hashing for large-scale unsupervised cross-modal retrieval,” inProceed- ings of the IEEE/CVF international conference on computer vision, 2019, pp. 3027–3035

    work page 2019

  12. [12]

    Unsupervised generative adversarial cross-modal hashing,

    J. Zhang, Y . Peng, and M. Yuan, “Unsupervised generative adversarial cross-modal hashing,” inProceedings of the AAAI conference on artifi- cial intelligence, vol. 32, no. 1, 2018

    work page 2018

  13. [13]

    Generative adversarial networks: An overview,

    A. Creswell, T. White, V . Dumoulin, K. Arulkumaran, B. Sengupta, and A. A. Bharath, “Generative adversarial networks: An overview,”IEEE signal processing magazine, vol. 35, no. 1, pp. 53–65, 2018

    work page 2018

  14. [14]

    Creating something from nothing: Unsupervised knowledge distillation for cross-modal hashing,

    H. Hu, L. Xie, R. Hong, and Q. Tian, “Creating something from nothing: Unsupervised knowledge distillation for cross-modal hashing,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 3123–3132

    work page 2020

  15. [15]

    Deep graph-neighbor coherence preserving network for unsupervised cross-modal hashing,

    J. Yu, H. Zhou, Y . Zhan, and D. Tao, “Deep graph-neighbor coherence preserving network for unsupervised cross-modal hashing,” inProceed- ings of the AAAI conference on artificial intelligence, vol. 35, no. 5, 2021, pp. 4626–4634

    work page 2021

  16. [16]

    Unsupervised contrastive cross-modal hashing,

    P. Hu, H. Zhu, J. Lin, D. Peng, Y .-P. Zhao, and X. Peng, “Unsupervised contrastive cross-modal hashing,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 3, pp. 3877–3889, 2023

    work page 2023

  17. [17]

    When clip meets cross- modal hashing retrieval: A new strong baseline,

    X. Xia, G. Dong, F. Li, L. Zhu, and X. Ying, “When clip meets cross- modal hashing retrieval: A new strong baseline,”Information Fusion, vol. 100, p. 101968, 2023

    work page 2023

  18. [18]

    Self-attentive clip hashing for unsupervised cross-modal retrieval,

    H. Yu, S. Ding, L. Li, and J. Wu, “Self-attentive clip hashing for unsupervised cross-modal retrieval,” inProceedings of the 4th ACM International Conference on Multimedia in Asia, 2022, pp. 1–7

    work page 2022

  19. [19]

    Unsupervised hashing retrieval via efficient correlation distillation,

    Z. Xi, X. Wang, and P. Cheng, “Unsupervised hashing retrieval via efficient correlation distillation,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, no. 7, pp. 3529–3541, 2023

    work page 2023

  20. [20]

    Work together: Correlation-identity reconstruction hashing for unsupervised cross-modal retrieval,

    L. Zhu, X. Wu, J. Li, Z. Zhang, W. Guan, and H. T. Shen, “Work together: Correlation-identity reconstruction hashing for unsupervised cross-modal retrieval,”IEEE Transactions on Knowledge and Data Engineering, vol. 35, no. 9, pp. 8838–8851, 2022

    work page 2022

  21. [21]

    Clip4hashing: unsupervised deep hashing for cross-modal video-text retrieval,

    Y . Zhuo, Y . Li, J. Hsiao, C. Ho, and B. Li, “Clip4hashing: unsupervised deep hashing for cross-modal video-text retrieval,” inProceedings of the 2022 international conference on multimedia retrieval, 2022, pp. 158–166

    work page 2022

  22. [22]

    Dual self- paced cross-modal hashing,

    Y . Sun, J. Dai, Z. Ren, Y . Chen, D. Peng, and P. Hu, “Dual self- paced cross-modal hashing,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 14, 2024, pp. 15 184–15 192

    work page 2024

  23. [23]

    Clip-based adaptive graph attention network for large-scale unsupervised multi-modal hashing retrieval,

    Y . Li, M. Ge, M. Li, T. Li, and S. Xiang, “Clip-based adaptive graph attention network for large-scale unsupervised multi-modal hashing retrieval,”Sensors, vol. 23, no. 7, p. 3439, 2023

    work page 2023

  24. [24]

    Clip-based fusion-modal reconstructing hashing for large-scale unsupervised cross- modal retrieval,

    L. Mingyong, L. Yewen, G. Mingyuan, and M. Longfei, “Clip-based fusion-modal reconstructing hashing for large-scale unsupervised cross- modal retrieval,”International Journal of Multimedia Information Re- trieval, vol. 12, no. 1, p. 2, 2023

    work page 2023

  25. [25]

    Cross-domain transfer hashing for efficient cross-modal retrieval,

    F. Li, B. Wang, L. Zhu, J. Li, Z. Zhang, and X. Chang, “Cross-domain transfer hashing for efficient cross-modal retrieval,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 10, pp. 9664– 9677, 2024

    work page 2024

  26. [26]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clarket al., “Learning transferable visual models from natural language supervision,” inInternational conference on machine learning. PMLR, 2021, pp. 8748–8763

    work page 2021

  27. [27]

    Remarks on some nonpara- metric estimates of a density function,

    R. A. Davis, K.-S. Lii, and D. N. Politis, “Remarks on some nonpara- metric estimates of a density function,” inSelected Works of Murray Rosenblatt. Springer, 2011, pp. 95–100

    work page 2011

  28. [28]

    On estimation of a probability density function and mode,

    E. Parzen, “On estimation of a probability density function and mode,” The annals of mathematical statistics, vol. 33, no. 3, pp. 1065–1076, 1962

    work page 1962

  29. [29]

    Unsupervised Data-Efficient Cross-Modal Retrieval with Global-Neighborhood Alignment Hashing

    R. Li, X. Ma, Z. Weng, Y . Zhang, G. Luo, H. Zhuang, Z. Lin, and Y .- P. Tan, “Unsupervised data-efficient cross-modal retrieval with global- neighborhood alignment hashing,”arXiv preprint arXiv:2606.31517, 2026. 13

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  30. [30]

    Cross-modality binary code learning via fusion similarity hashing,

    H. Liu, R. Ji, Y . Wu, F. Huang, and B. Zhang, “Cross-modality binary code learning via fusion similarity hashing,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 7380– 7388

    work page 2017

  31. [31]

    The Power of Scale for Parameter-Efficient Prompt Tuning

    B. Lester, R. Al-Rfou, and N. Constant, “The power of scale for parameter-efficient prompt tuning,”arXiv preprint arXiv:2104.08691, 2021

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  32. [32]

    Learning to prompt for vision- language models,

    K. Zhou, J. Yang, C. C. Loy, and Z. Liu, “Learning to prompt for vision- language models,”International Journal of Computer Vision, vol. 130, no. 9, pp. 2337–2348, 2022

    work page 2022

  33. [33]

    Tcp: Textual-based class-aware prompt tuning for visual-language model,

    H. Yao, R. Zhang, and C. Xu, “Tcp: Textual-based class-aware prompt tuning for visual-language model,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 23 438–23 448

    work page 2024

  34. [34]

    Class-specific prompt learning for vision–language models,

    R. Li, Y . Chen, Z. Weng, Z. Lin, and Y .-P. Tan, “Class-specific prompt learning for vision–language models,”IEEE Transactions on Neural Networks and Learning Systems, 2025

    work page 2025

  35. [35]

    Class-specific knowledge-guided multimodal prompt tuning for few-shot class-incremental learning,

    F. Xiong, Z. Yuan, X. Wu, and C. Xu, “Class-specific knowledge-guided multimodal prompt tuning for few-shot class-incremental learning,” IEEE Transactions on Circuits and Systems for Video Technology, 2025

    work page 2025

  36. [36]

    Map-vla: Memory-augmented prompting for vision-language-action model in robotic manipulation.arXiv preprint arXiv:2511.09516,

    R. Li, W. Guo, Z. Wu, C. Wang, H. Deng, Z. Weng, Y .-P. Tan, and Z. Wang, “Map-vla: Memory-augmented prompting for vision-language- action model in robotic manipulation,”arXiv preprint arXiv:2511.09516, 2025

    work page arXiv 2025

  37. [37]

    Kernel methods in machine learning,

    T. Hofmann, B. Sch ¨olkopf, and A. J. Smola, “Kernel methods in machine learning,”The Annals of Statistics, vol. 36, no. 3, pp. 1171 – 1220, 2008

    work page 2008

  38. [38]

    Kernelized locality-sensitive hashing for scalable image search,

    B. Kulis and K. Grauman, “Kernelized locality-sensitive hashing for scalable image search,” in2009 IEEE 12th international conference on computer vision. IEEE, 2009, pp. 2130–2137

    work page 2009

  39. [39]

    Supervised hashing with kernels,

    W. Liu, J. Wang, R. Ji, Y .-G. Jiang, and S.-F. Chang, “Supervised hashing with kernels,” in2012 IEEE conference on computer vision and pattern recognition. IEEE, 2012, pp. 2074–2081

    work page 2012

  40. [40]

    Label-consistent kernel transform learning-based sparse hashing for cross-modal retrieval,

    J. Maggu, A. Goel, and R. Kumar, “Label-consistent kernel transform learning-based sparse hashing for cross-modal retrieval,”Knowledge and Information Systems, vol. 67, no. 8, pp. 6937–6967, 2025

    work page 2025

  41. [41]

    A simple framework for contrastive learning of visual representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” inInternational conference on machine learning. PMLR, 2020, pp. 1597–1607

    work page 2020

  42. [42]

    A survey on learning to hash,

    J. Wang, T. Zhang, N. Sebe, H. T. Shenet al., “A survey on learning to hash,”IEEE transactions on pattern analysis and machine intelligence, vol. 40, no. 4, pp. 769–790, 2017

    work page 2017

  43. [43]

    The mir flickr retrieval evaluation,

    M. J. Huiskes and M. S. Lew, “The mir flickr retrieval evaluation,” in Proceedings of the 1st ACM international conference on Multimedia information retrieval, 2008, pp. 39–43

    work page 2008

  44. [44]

    Nus-wide: a real-world web image database from national university of singapore,

    T.-S. Chua, J. Tang, R. Hong, H. Li, Z. Luo, and Y . Zheng, “Nus-wide: a real-world web image database from national university of singapore,” inProceedings of the ACM international conference on image and video retrieval, 2009, pp. 1–9

    work page 2009

  45. [45]

    Collecting image annotations using amazon’s mechanical turk,

    C. Rashtchian, P. Young, M. Hodosh, and J. Hockenmaier, “Collecting image annotations using amazon’s mechanical turk,” inProceedings of the NAACL HLT 2010 workshop on creating speech and language data with Amazon’s Mechanical Turk, 2010, pp. 139–147

    work page 2010

  46. [46]

    A new approach to cross-modal multime- dia retrieval,

    N. Rasiwasia, J. Costa Pereira, E. Coviello, G. Doyle, G. R. Lanckriet, R. Levy, and N. Vasconcelos, “A new approach to cross-modal multime- dia retrieval,” inProceedings of the 18th ACM international conference on Multimedia, 2010, pp. 251–260

    work page 2010

  47. [47]

    Cross-modal retrieval with correspondence autoencoder,

    F. Feng, X. Wang, and R. Li, “Cross-modal retrieval with correspondence autoencoder,” inProceedings of the 22nd ACM International Conference on Multimedia, ser. MM ’14. New York, NY , USA: Association for Computing Machinery, 2014, p. 7–16. [Online]. Available: https://doi.org/10.1145/2647868.2654902

    work page doi:10.1145/2647868.2654902 2014

  48. [48]

    Deep supervised cross-modal retrieval,

    L. Zhen, P. Hu, X. Wang, and D. Peng, “Deep supervised cross-modal retrieval,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 10 394–10 403

    work page 2019

  49. [49]

    Deep unsupervised hashing via external guidance,

    Q. Song, X. Liu, H. Zhu, J. T. Zhou, X. Peng, and P. Hu, “Deep unsupervised hashing via external guidance,” inProceedings of the 42nd International Conference on Machine Learning, ser. Proceedings of Machine Learning Research, A. Singh, M. Fazel, D. Hsu, S. Lacoste- Julien, F. Berkenkamp, T. Maharaj, K. Wagstaff, and J. Zhu, Eds., vol

    work page

  50. [50]

    56 414–56 428

    PMLR, 13–19 Jul 2025, pp. 56 414–56 428. [Online]. Available: https://proceedings.mlr.press/v267/song25h.html

    work page 2025

  51. [51]

    Adam: A Method for Stochastic Optimization

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  52. [52]

    Learning to predict visual attributes in the wild,

    K. Pham, K. Kafle, Z. Lin, Z. Ding, S. Cohen, Q. Tran, and A. Shri- vastava, “Learning to predict visual attributes in the wild,” in2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 13 013–13 023

    work page 2021