Pith. sign in

REVIEW 6 major objections 8 minor 52 references

KALAHash: Knowledge-Anchored Low-Resource Adaptation for Deep Hashing

T0 review · 6 major / 8 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Class-name text embeddings serve as anchors that let deep hashing adapt from as few as one image per class.

desk verdict Promising CLoRA idea and a broad empirical study, but the KIDDO loss as written is dimensionally undefined, which undercuts the paper's central contribution. read the letter →

arxiv 2412.19417 v1 pith:5NJCT4WE submitted 2024-12-27 cs.CV

Shu Zhao , Tan Yu , Xiaoshuai Hao , Wenchao Ma , Vijaykrishnan Narayanan This is my paper · ORCID
classification cs.CV
keywords deephashinglow-resourceadaptationparameter-efficientfine-tuningLoRAvision-languagemodelsCLIPembeddingsfew-shotretrievaldiscreteoptimization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper takes on a setting most deep-hashing methods ignore: adapting a pre-trained model to a new retrieval task when only one to eight labeled images per class are available. It claims that in this regime, existing methods degrade badly because the few samples shift the learned hash-code distribution away from what the pre-trained model knows. To fix this, KALAHash treats the CLIP text embedding of each category name as an anchor point, injecting that knowledge into both the fine-tuning adapter and the hash-code optimization. Concretely, CLoRA builds LoRA's low-rank update from class-level text embeddings, and KIDDO adds an alignment loss that pushes binary codes to reproduce the one-hot label matrix through the text knowledge. The paper reports that the combination beats eight deep-hashing baselines on CIFAR-10, NUS-WIDE, and MS-COCO at 1 to 8 shots and claims a 4x data efficiency.

What carries the argument

The carrying mechanism is a pair of knowledge-injection devices built on the CLIP text encoder. CLoRA constrains a LoRA weight update to the span of class text embeddings: for each input, it averages the vision tokens, picks the top text embeddings by cosine similarity, and forms $\Delta W = \eta \sum_i \hat{k}_i q_i^T$, so the adapter can only move weights along semantic directions the category names define. KIDDO adds an alignment loss $\|Y - T^\top B\|_2^2$ to the standard pairwise similarity and quantization losses and solves for binary codes $B$ column by column with discrete cyclic coordinate descent, injecting the same textual knowledge into the binary codes themselves.

What would settle it

Take a 1-shot retrieval benchmark from a domain CLIP was not trained on, such as medical or satellite imagery, and compare KALAHash against plain LoRA on the same frozen CLIP backbone; if mAP does not improve, the text-anchor assumption fails. A cleaner controlled test: permute the category-name text embeddings randomly before training, and if retrieval mAP stays high after permuting, the semantic content of the anchors is not what carries the result.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that class-level textual knowledge from a pre-trained vision-language model can stand in for the missing visual variety in low-resource deep hashing. The hash function is anchored by category-name embeddings: CLoRA replaces the random left factor of a LoRA update with a text embedding selected by cosine similarity to the pooled visual tokens, and KIDDO aligns the learned binary codes with the same text knowledge while optimizing them with discrete cyclic coordinate descent. The paper's experiments report consistent gains over eight baselines at 1, 2, 4, and 8 shots, with the largest margins in the 1-shot case, and it shows CLoRA can be dropped onto existing hashing losses to improve them.

Load-bearing premise

The load-bearing premise is that the CLIP text embedding of a category name is a faithful anchor for that category's visual hash codes, so if text and vision embeddings do not line up for a target domain, both CLoRA's anchor selection and KIDDO's alignment loss lose their grounding.

Editorial extensions

If this is right

  • At the 1-shot and 2-shot settings, KALAHash reports mAP gains over all eight baselines on CIFAR-10, NUS-WIDE, and MS-COCO, with the largest margins at 1 shot.
  • The plug-and-play CLoRA component improves mAP for every baseline it is inserted into, with gains up to 12.52 mAP points on CIFAR-10.
  • Ablation results indicate both components matter: removing CLoRA costs 2.38 to 11.16 mAP points, and removing KIDDO costs 3.72 to 6.65 mAP points across the three datasets.
  • When the number of training samples grows to 500 shots on CIFAR-10, KALAHash remains competitive with state-of-the-art methods, suggesting the textual anchor does not harm performance once data is plentiful.
  • CLoRA adds negligible inference overhead across tested backbones, with measured increases of 0.01 to 0.05 milliseconds per image.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper: the anchor construction should transfer to any retrieval head built on CLIP features, not just hash layers, because it only changes how adapter weights are formed.
  • Beyond the paper: if anchor selection is the true source of gain, replacing single category-name prompts with richer per-class descriptions should help further on fine-grained benchmarks, which the paper does not test.
  • Beyond the paper: the top-1 cosine anchor selection in Equation (6) could misfire for visually similar classes, so a soft mixture of anchors or a learned selection rule is a natural stress test of that assumption.
  • Beyond the paper: the 4x data-efficiency claim suggests a direct testable rule, namely that KALAHash at one shot should match the retrieval mAP of existing methods at roughly four shots, which can be checked against the scaling curves in Figure 4.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

6 major / 8 minor

Summary. The paper introduces KALAHash, a method for low-resource (1-8 shots per class) adaptation of deep hashing models. It builds on CLIP: CLoRA constructs low-rank adaptation matrices for the vision transformer using top-r class text embeddings as anchors, and KIDDO adds a discrete optimization objective that aligns binary codes with text embeddings. Experiments on CIFAR-10, NUS-WIDE, and MS-COCO report consistent mAP improvements over eight hashing baselines, a plug-and-play study adding CLoRA to baselines, ablations, scaling analyses, and inference-time measurements. The central claims are that KALAHash significantly boosts retrieval performance and achieves 4x data efficiency.

Significance. If the empirical claims were supported, the paper would make a useful contribution: it is among the first to address low-resource adaptation specifically for deep hashing, and the idea of using class-level text embeddings both as a parameter-efficient adaptation mechanism (CLoRA) and as optimization anchors (KIDDO) is practical and reasonably novel. The authors provide a public code link and extensive experiments across three datasets, several backbones, and multiple shot settings, which is a strength. However, the current manuscript contains serious inconsistencies in the formulation of KIDDO, inconsistent baseline numbers across tables, missing statistical validation, and an unsubstantiated 4x data-efficiency claim, so the central claims are not yet supported as written.

major comments (6)
  1. [Knowledge-Guided Discrete Optimization, Eq. (9)] The alignment loss La = ||Y - T^T B||^2 is dimensionally inconsistent. With Y in R^{C x n}, B in {±1}^{n x b}, and T = G(K) with K in R^{C x d_t}, if G outputs b-dimensional rows then T is C x b, so T^T B is (b x C)(n x b), which is not conformable unless C = n. The expression T B^T, giving C x n, would be conformable, suggesting a missing transpose. Since Eq. (11) and the DCC update in Eq. (12) inherit this mismatch, the derivation cannot be verified, and the formula for S and the update Bi = sign(S_i - B'^T T'_i) contain further dimensional inconsistencies. This is load-bearing because KIDDO is a central contribution and the ablation in Table 3 attributes 2.38-11.16 mAP points to it. Please correct the objective, re-derive the DCC update, and verify the equations.
  2. [Knowledge-Guided Discrete Optimization, Eq. (9)] Y is defined as a matrix of one-hot label vectors, yet the experiments on NUS-WIDE and MS-COCO treat these as multi-label datasets, and the evaluation protocol states that two images are similar if they share at least one label. With a one-hot Y, the alignment loss cannot represent co-occurring labels, so the reported multi-label gains do not follow from the described objective. Please specify whether Y is multi-hot for these datasets and ensure the equations and implementation match this choice.
  3. [Tables 1 and 2] The baseline numbers are inconsistent between the two tables. For MS-COCO 1-shot, HSWD is 60.15 in Table 1 but 58.55 in Table 2; for CIFAR-10 1-shot, OrthoHash is 46.68 in Table 1 but 49.50 in Table 2. This inconsistency undermines the plug-and-play gains in Table 2, since the improvements may be computed against different baseline values. Please unify the tables and re-report the results consistently.
  4. [Experiments (overall)] None of the reported results include error bars, standard deviations, or multiple random splits. In 1-shot to 8-shot settings, the training set selection is stochastic and variance is typically high; without repeated runs, the claim that KALAHash "significantly boosts" retrieval performance is not statistically supported. Please report mean ± std across at least 3 seeds (or provide confidence intervals) for the main tables and for the plug-and-play results.
  5. [Abstract and Experiments] The claim of "4x data efficiency" is not quantified anywhere in the manuscript. There is no experiment showing that KALAHash at k shots matches or exceeds baselines at 4k shots, and no explicit definition of the 4x factor. Please add a clear analysis, e.g., using Figure 4 or a dedicated table, to substantiate this claim, or remove it from the abstract and introduction.
  6. [Implementation Details and Comparison Protocol] All baselines are trained without the text embeddings (CLIP class-name knowledge) that KALAHash uses in both CLoRA and KIDDO. The statement that "all methods, including baselines, use the same backbone model, optimizer, training hyperparameters, etc." is therefore incomplete as a fairness argument, because the proposed method has access to additional semantic side information. Since the method's gains could stem primarily from this extra prior, please add a baseline that also uses text knowledge (e.g., a CLIP zero-shot hash adapted variant, or a variant with text features appended) or explicitly discuss this limitation.
minor comments (8)
  1. [Class-Level Textual Knowledge Generation] The text contains typos such as "konwledge" for "knowledge"; please proofread the manuscript.
  2. [Eq. (8)] In the similarity loss Ls, it is not explicitly stated whether H or B is used in θij; please clarify that H is the real-valued hash output before binarization.
  3. [Eq. (12)] The definitions of S, S', T', and B' are ambiguous and contain possible notation errors (e.g., "S = βYT + γH" appears to lack a transpose and is dimensionally unclear). The formulas need to be rewritten consistently, even after correcting Eq. (9).
  4. [Figure 1] The caption says "mean Average Precision scores" but should use "mean Average Precision (mAP)" for consistency with the tables.
  5. [Introduction] The text states "FFT achieves a Silhouette Score of 50.0%" without noting that the raw score is normalized from [-1,1] to [0,100]; please add that clarification in the main text as well as the caption.
  6. [Table 5] The notation "w.o.CLoRA" and "w.CLoRA" is unusual; please spell out "without" and "with" in the table caption or use standard abbreviations.
  7. [Appendix: Scalability of the Number of Bits] The sentence "As the number of training samples increases, our approach consistently improves the retrieval performance" should read "As the number of bits increases" based on the figure and context.
  8. [Appendix Tables 8 and 9] Tables 8 and 9 are not explicitly referenced in the main text; please add cross-references or integrate this content into the main experimental section.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the text-knowledge anchors are external inputs, and the reported retrieval gains are evaluated on held-out labels, so the derivation is self-contained.

full rationale

The paper's derivation chain does not reduce to its own inputs by construction. CLoRA (Eq. 5) and KIDDO (Eq. 9) use CLIP text embeddings K and the label matrix Y as externally specified side information; K is pre-extracted once from category names and is not fitted to the retrieval metric, and the evaluation protocol randomly splits per-class training samples while computing mAP on separate held-out query/gallery sets. The reported improvements are empirical comparisons against baselines under the same low-resource protocol, so the central claim does not rename a fitted quantity as a prediction. The paper's self-citations appear only in related work or as general context (e.g., prior hashing losses, low-resource adaptation references) and are not load-bearing for the claimed derivation. Concerns that baselines do not receive the same text-side information, or that Eq. (9) has a dimension mismatch and uses one-hot labels on multi-label data, are correctness and experimental-fairness issues rather than evidence that a prediction is equivalent to its input by definition.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the alignment of CLIP text and vision spaces, on the query selecting a single correct text anchor per image, and on hand-set loss weights. No new physical entities are introduced; CLoRA and KIDDO are algorithmic components. The one-hot label assumption is also an axiom that conflicts with the multi-label datasets.

free parameters (4)
  • Loss weights alpha, beta, gamma = 0.1, 1.0, 3.0
    Hand-set in Implementation Details; only gamma sensitivity is reported, and the same values are reused across all datasets and shot settings.
  • LoRA rank r = 1
    Default rank; with r=1 the CLoRA update is a single text-anchored rank-1 direction, and sensitivity to r is not reported.
  • LoRA scale eta = 1.0
    Fixed scale in Eq. (7); no sensitivity analysis is provided.
  • CLoRA insertion position = key and value matrices of the last transformer layer
    One of many possible insertion positions; Figure 7 shows a sensitivity grid, but this default is used in all main experiments.
assumptions (4)
  • domain assumption CLIP text embeddings of category names are aligned with CLIP visual features well enough to serve as fixed anchors during low-shot adaptation.
    CLoRA replaces the learned left LoRA factor with mapped text embeddings (Eq. 5), and KIDDO uses T^T B to approximate labels (Eq. 9).
  • domain assumption With one training image per class, the cosine-similarity selection in Eq. (6) picks the correct class-level knowledge vector for each input.
    The Top-r selection uses avg(V) without label conditioning, and r=1 means the entire update depends on one text anchor being right.
  • standard math Discrete cyclic coordinate descent from SDH (Shen et al. 2015) yields the column-wise optimum of the mixed binary optimization in Eq. (11).
    The paper adopts DCC and writes Eq. (12) as the update; the update's dimensions are not fully consistent, so the derivation leans on the original method.
  • ad hoc to paper The label matrix Y can be treated as one-hot even on the multi-label datasets NUS-WIDE and MS-COCO.
    Eq. (9) calls Y one-hot, but NUS-WIDE and MS-COCO are multi-label; unless Y is actually multi-hot, the alignment loss is misspecified on two of the three main benchmarks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of KALAHash: Knowledge-Anchored Low-Resource Adaptation for Deep Hashing." pith.science (2026). https://pith.science/paper/5NJCT4WE

@misc{pith2026241219417,
  author       = {Pith},
  title        = {Pith review of: KALAHash: Knowledge-Anchored Low-Resource Adaptation for Deep Hashing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5NJCT4WE}},
  note         = {Machine review of arXiv:2412.19417}
}
read the original abstract

Deep hashing has been widely used for large-scale approximate nearest neighbor search due to its storage and search efficiency. However, existing deep hashing methods predominantly rely on abundant training data, leaving the more challenging scenario of low-resource adaptation for deep hashing relatively underexplored. This setting involves adapting pre-trained models to downstream tasks with only an extremely small number of training samples available. Our preliminary benchmarks reveal that current methods suffer significant performance degradation due to the distribution shift caused by limited training samples. To address these challenges, we introduce Class-Calibration LoRA (CLoRA), a novel plug-and-play approach that dynamically constructs low-rank adaptation matrices by leveraging class-level textual knowledge embeddings. CLoRA effectively incorporates prior class knowledge as anchors, enabling parameter-efficient fine-tuning while maintaining the original data distribution. Furthermore, we propose Knowledge-Guided Discrete Optimization (KIDDO), a framework to utilize class knowledge to compensate for the scarcity of visual information and enhance the discriminability of hash codes. Extensive experiments demonstrate that our proposed method, Knowledge- Anchored Low-Resource Adaptation Hashing (KALAHash), significantly boosts retrieval performance and achieves a 4x data efficiency in low-resource scenarios.

Figures

Figures reproduced from arXiv: 2412.19417 by the authors.

Figure 1
Figure 1. Performance comparison in low-resource set [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Architecture overview of the proposed KALAHash [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Architecture of the proposed CLoRA module. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Performance comparison of KALAHash and base [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 6
Figure 6. Figure 6: t-SNE visualization of learned hash codes for Full Fine-Tuning (FFT), Lock Backbone (LB), and our proposed method [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Parameter sensitivity analysis for KALAHash, showing mAP performance across different hyper-parameter settings. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Performance of KALAHash as the number of bits [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Sensitivity analysis of KALAHash with respect to [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]
Figure 10
Figure 10. Figure 10: Precision-Recall curves for KALAHash and baseline methods on NUS-WIDE, MS-COCO, and CIFAR-10 datasets. [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 45 canonical work pages

  1. [1]

    Cao, Y.; Long, M.; Liu, B.; and Wang, J. 2018. Deep Cauchy Hashing for Hamming Space Retrieval. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    work page 2018

  2. [2]

    Cao, Z.; Long, M.; Wang, J.; and Yu, P. S. 2017. HashNet: Deep Learning to Hash by Continuation. In International Conference on Computer Vision (ICCV)

    work page 2017

  3. [3]

    Chua, T.; Tang, J.; Hong, R.; Li, H.; Luo, Z.; and Zheng, Y. 2009. NUS-WIDE: a real-world web image database from National University of Singapore. In Conference On Image And Video Retrieval (CIVR)

    work page 2009

  4. [4]

    This Is My Unicorn, Fluffy

    Cohen, N.; Gal, R.; Meirom, E. A.; Chechik, G.; and Atzmon, Y. 2022. "This Is My Unicorn, Fluffy": Personalizing Frozen Vision-Language Representations. In European Conference on Computer Vision (ECCV)

    work page 2022

  5. [5]

    D.; Yang, P.; and Li, P

    Doan, K. D.; Yang, P.; and Li, P. 2022. One Loss for Quantization: Deep Hashing with Discrete Wasserstein Distributional Matching. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    work page 2022

  6. [6]

    Dosovitskiy, A.; Beyer, L.; Kolesnikov, A.; Weissenborn, D.; Zhai, X.; Unterthiner, T.; Dehghani, M.; Minderer, M.; Heigold, G.; Gelly, S.; Uszkoreit, J.; and Houlsby, N. 2021. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. In International Conference on Learning Representations (ICLR)

    work page 2021

  7. [7]

    Finn, C.; Abbeel, P.; and Levine, S. 2017. Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks. In International Conference on Machine Learning (ICML)

    work page 2017

  8. [8]

    Gui, L.; Wang, Y.; and Hebert, M. 2017. Few-Shot Hash Learning for Image Retrieval. In International Conference on Computer Vision Workshops (ICCV Workshops)

    work page 2017

Show all 52 references
  1. [9]

    Hao, X.; Li, R.; Zhang, H.; Li, D.; Yin, R.; Jung, S.; Park, S.; Yoo, B.; Zhao, H.; and Zhang, J. 2024 a . MapDistill: Boosting Efficient Camera-Based HD Map Construction via Camera-LiDAR Fusion Model Distillation. In European Conference on Computer Vision (ECCV)

    2024

  2. [10]

    Hao, X.; Wei, M.; Yang, Y.; Zhao, H.; Zhang, H.; Zhou, Y.; Wang, Q.; Li, W.; Kong, L.; and Zhang, J. 2024 b . Is Your HD Map Constructor Reliable under Sensor Corruptions? In Conference on Neural Information Processing Systems (NeurIPS)

    2024

  3. [11]

    Hao, X.; and Zhang, W. 2023. Uncertainty-Aware Alignment Network for Cross-Domain Video-Text Retrieval. In Conference on Neural Information Processing Systems (NeurIPS)

    2023

  4. [12]

    Hao, X.; Zhang, W.; Wu, D.; Zhu, F.; and Li, B. 2023. Dual Alignment Unsupervised Domain Adaptation for Video-Text Retrieval. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    2023

  5. [13]

    A.; and Sclaroff, S

    He, K.; C akir, F.; Bargal, S. A.; and Sclaroff, S. 2018. Hashing as Tie-Aware Learning to Rank. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    2018

  6. [14]

    He, K.; Zhang, X.; Ren, S.; and Sun, J. 2016. Deep Residual Learning for Image Recognition. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    2016

  7. [15]

    T.; Ng, K

    Hoe, J. T.; Ng, K. W.; Zhang, T.; Chan, C. S.; Song, Y.; and Xiang, T. 2021. One Loss for All: Deep Hashing with a Single Cosine Similarity based Learning Objective. In Conference on Neural Information Processing Systems (NeurIPS)

    2021

  8. [16]

    J.; Shen, Y.; Wallis, P.; Allen - Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; and Chen, W

    Hu, E. J.; Shen, Y.; Wallis, P.; Allen - Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; and Chen, W. 2022. LoRA: Low-Rank Adaptation of Large Language Models. In International Conference on Learning Representations (ICLR)

    2022

  9. [17]

    Jiang, Q.; and Li, W. 2018. Asymmetric Deep Supervised Hashing. In AAAI Conference on Artificial Intelligence (AAAI)

    2018

  10. [18]

    Krizhevsky, A.; and Hinton, G. 2009. Learning multiple layers of features from tiny images. Technical report, University of Toronto

    2009

  11. [19]

    Lai, H.; Pan, Y.; Liu, Y.; and Yan, S. 2015. Simultaneous feature learning and hash coding with deep neural networks. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    2015

  12. [20]

    Li, Q.; Sun, Z.; He, R.; and Tan, T. 2017. Deep Supervised Discrete Hashing. In Conference on Neural Information Processing Systems (NeurIPS)

    2017

  13. [21]

    Li, W.; Wang, S.; and Kang, W. 2016. Feature Learning Based Deep Supervised Hashing with Pairwise Labels. In International Joint Conference on Artificial Intelligence (IJCAI)

    2016

  14. [22]

    J.; Hays, J.; Perona, P.; Ramanan, D.; Doll \' a r, P.; and Zitnick, C

    Lin, T.; Maire, M.; Belongie, S. J.; Hays, J.; Perona, P.; Ramanan, D.; Doll \' a r, P.; and Zitnick, C. L. 2014. Microsoft COCO: Common Objects in Context. In European Conference on Computer Vision (ECCV)

    2014

  15. [23]

    Liu, H.; Li, C.; Wu, Q.; and Lee, Y. J. 2023. Visual Instruction Tuning. In Conference on Neural Information Processing Systems (NeurIPS)

    2023

  16. [24]

    Luo, X.; Wang, H.; Wu, D.; Chen, C.; Deng, M.; Huang, J.; and Hua, X. 2023. A Survey on Deep Hashing Methods. ACM Transactions on Knowledge Discovery from Data (TKDD)

    2023

  17. [25]

    A.; and Xie, S

    Mu, N.; Kirillov, A.; Wagner, D. A.; and Xie, S. 2022. SLIP: Self-supervision Meets Language-Image Pre-training. In European Conference on Computer Vision (ECCV)

    2022

  18. [26]

    W.; Zhu, X.; Song, Y.; and Xiang, T

    Ng, K. W.; Zhu, X.; Song, Y.; and Xiang, T. 2024. ConceptHash: Interpretable Fine-Grained Hashing via Concept Discovery. In IEEE/CVF Computer Vision and Pattern Recognition Conference Workshops (CVPRW)

    2024

  19. [27]

    Pan, Z.; Cai, J.; and Zhuang, B. 2023. Stitchable neural networks. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    2023

  20. [28]

    W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; Krueger, G.; and Sutskever, I

    Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; Krueger, G.; and Sutskever, I. 2021. Learning Transferable Visual Models From Natural Language Supervision. In International Conference on Machine Learning (ICML)

    2021

  21. [29]

    Rousseeuw, P. J. 1987. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. Journal of computational and applied mathematics

    1987

  22. [30]

    Shen, F.; Gao, X.; Liu, L.; Yang, Y.; and Shen, H. T. 2017. Deep Asymmetric Pairwise Hashing. In ACM International Conference on Multimedia (ACM MM)

    2017

  23. [31]

    Shen, F.; Shen, C.; Liu, W.; and Shen, H. T. 2015. Supervised Discrete Hashing. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    2015

  24. [32]

    Snell, J.; Swersky, K.; and Zemel, R. S. 2017. Prototypical Networks for Few-shot Learning. In Conference on Neural Information Processing Systems (NeurIPS)

    2017

  25. [33]

    Su, S.; Zhang, C.; Han, K.; and Tian, Y. 2018. Greedy Hash: Towards Fast Optimization for Accurate Hash Coding in CNN . In Conference on Neural Information Processing Systems (NeurIPS)

    2018

  26. [34]

    Van der Maaten, L.; and Hinton, G. 2008. Visualizing data using t-SNE. Journal of machine learning research (JMLR)

    2008

  27. [35]

    N.; Kaiser, L.; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, L.; and Polosukhin, I. 2017. Attention is All you Need. In Conference on Neural Information Processing Systems (NeurIPS)

    2017

  28. [36]

    Venkateswara, H.; Eusebio, J.; Chakraborty, S.; and Panchanathan, S. 2017. Deep Hashing Network for Unsupervised Domain Adaptation. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    2017

  29. [37]

    Wang, L.; Pan, Y.; Liu, C.; Lai, H.; Yin, J.; and Liu, Y. 2023 a . Deep Hashing with Minimal-Distance-Separated Hash Centers. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    2023

  30. [38]

    Wang, Q.; Yang, X.; Lin, S.; and Geng, X. 2023 b . Learngene: Inheriting Condensed Knowledge from the Ancestry Model to Descendant Models. CoRR, abs/2305.02279

    2023 arXiv

  31. [39]

    Wang, X.; Shi, Y.; and Kitani, K. M. 2016. Deep Supervised Hashing with Triplet Labels. In Asian Conference on Computer Vision (ACCV)

    2016

  32. [40]

    Wu, D.; Su, Q.; Li, B.; and Wang, W. 2024. Pairwise-Label-Based Deep Incremental Hashing with Simultaneous Code Expansion. In AAAI Conference on Artificial Intelligence (AAAI)

    2024

  33. [41]

    Xia, R.; Pan, Y.; Lai, H.; Liu, C.; and Yan, S. 2014. Supervised Hashing for Image Retrieval via Image Representation Learning. In AAAI Conference on Artificial Intelligence (AAAI)

    2014

  34. [42]

    Yang, E.; Wang, Z.; Shen, L.; Liu, S.; Guo, G.; Wang, X.; and Tao, D. 2024. AdaMerging: Adaptive Model Merging for Multi-Task Learning. In International Conference on Learning Representations (ICLR)

    2024

  35. [43]

    Yuan, L.; Wang, T.; Zhang, X.; Tay, F. E. H.; Jie, Z.; Liu, W.; and Feng, J. 2020. Central Similarity Quantization for Efficient Image and Video Retrieval. In IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR)

    2020

  36. [44]

    B.; Goldberg, Y.; and Ravfogel, S

    Zaken, E. B.; Goldberg, Y.; and Ravfogel, S. 2022. BitFit: Simple Parameter-efficient Fine-tuning for Transformer-based Masked Language-models. In Annual Meeting of the Association for Computational Linguistics (ACL)

    2022

  37. [45]

    Zhao, S.; Wu, D.; Zhang, W.; Zhou, Y.; Li, B.; and Wang, W. 2020. Asymmetric Deep Hashing for Efficient Hash Code Compression. In ACM International Conference on Multimedia (ACM MM)

    2020

  38. [46]

    Zhao, S.; Wu, D.; Zhou, Y.; Li, B.; and Wang, W. 2021. Rescuing Deep Hashing from Dead Bits Problem. In International Joint Conference on Artificial Intelligence (IJCAI)

    2021

  39. [47]

    Zhao, S.; and Xu, H. 2023 a . Less is More: Toward Zero-Shot Local Scene Graph Generation via Foundation Models. CoRR, abs/2310.01356

    2023 arXiv

  40. [48]

    Zhao, S.; and Xu, H. 2023 b . NEUCORE: Neural Concept Reasoning for Composed Image Retrieval. In UniReps, Proceedings of Machine Learning Research

    2023

  41. [49]

    Zhao, S.; Zou, X.; Yu, T.; and Xu, H. 2024. Reconstruct before Query: Continual Missing Modality Learning with Decomposed Prompt Collaboration. CoRR, abs/2403.11373

    2024 arXiv

  42. [50]

    C.; and Liu, Z

    Zhou, K.; Yang, J.; Loy, C. C.; and Liu, Z. 2022. Learning to Prompt for Vision-Language Models. International Journal of Computer Vision (IJCV)

    2022

  43. [51]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  44. [52]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.