Pith. sign in

REVIEW 4 major objections 6 minor 41 references

CacheFL: Privacy-Preserving and Efficient Federated Cache Model Fine-Tuning for Vision-Language Models

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read This paper argues that federated fine-tuning of CLIP can be reduced to updating a lightweight cache matrix, and that this cache-only approach beats prompt- and adapter-based federated baselines on 11 image-classification datasets.

desk verdict A useful empirical recipe for federated cache-model fine-tuning of CLIP, undermined by a convergence theorem that does not apply to the trained (non-convex) objective. read the letter →

arxiv 2505.05130 v2 pith:EYHBI37V submitted 2025-05-08 cs.DC

Mengjun Yi , Hanwen Zhang , Hui Dou , Jian Zhao , Furao Shen This is my paper · ORCID
classification cs.DC
keywords federatedlearningcachemodelvision-languagemodelsCLIPparameter-efficientfine-tuningnon-IIDdataprivacypreservationsyntheticinitialization
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

CacheFL is a federated fine-tuning method for vision-language models like CLIP that trains only a lightweight cache model instead of the full backbone. The cache stores visual features $W_1$ and one-hot labels $W_2$ built from a server-generated class-balanced synthetic dataset; clients update only $W_1$ on private data and the server aggregates updates with federated averaging. The paper argues this design preserves privacy, cuts communication and computation to a small fraction of full-model fine-tuning, and mitigates non-IID data through the class-balanced initialization. Experiments on 11 datasets report average gains of roughly 7 to 10 points over PromptFL, and a convergence theorem claims an $O(1/t)$ rate under standard assumptions.

What carries the argument

The central object is the cache model, composed of $W_1$, a $C \times NK$ matrix of CLIP image features from $NK$ synthetic images, and $W_2$, an $N \times NK$ matrix of one-hot labels for those images. Predictions combine the frozen CLIP logits $f_{\text{vision}} W_{\text{text}}^T$ with the cache logits $\exp(-\beta(1 - f_{\text{vision}} W_1)) W_2^T$, where the exponential term converts feature distance into similarity weights, so the cache acts as a retrievable class-balanced memory. During federated training only $W_1$ is updated locally by SGD on cross-entropy loss, with $W_2$ frozen, and the server aggregates client updates by federated averaging. Theorem 1 then bounds the objective gap by $O(1/t)$ under smoothness, strong convexity, and bounded-gradient assumptions.

What would settle it

Compute the smallest eigenvalue of the Hessian of the local cross-entropy loss with respect to $W_1$ at several points along a training run; if it is ever zero or negative, Assumption 2 fails and Theorem 1 does not apply to CacheFL as run.

Watch

Extended reading notes

Core claim

The central claim is that a frozen CLIP backbone plus a trainable cache model is a sufficient and superior adapter for federated fine-tuning. The cache model stores visual feature weights $W_1$ and frozen one-hot label weights $W_2$ built from class-balanced synthetic images; predictions fuse CLIP's text-image logits with a distance-weighted cache lookup, $\text{logits} = f_{\text{vision}} W_{\text{text}}^T + \alpha \exp(-\beta(1 - f_{\text{vision}} W_1)) W_2^T$, and only $W_1$ is trained locally and aggregated. The paper shows this object, initialized from synthetic data and then refined federated, outperforms FedAvg, FedProx, AdapterFL, PromptFL, and zero-shot CLIP across 11 datasets and three data-partition settings, with the largest margins in non-IID regimes, while sending about 100,000 parameters per round and resisting gradient-inversion attacks.

Load-bearing premise

The proof's claimed speed of convergence relies on every client's loss being strongly convex, but the actual loss includes an exponential distance term inside the cross-entropy, so strong convexity is asserted rather than proved and the bound does not follow if this assumption fails.

Editorial extensions

If this is right

  • Per-round communication drops to the size of the cache matrix, about 100,000 parameters, making federated fine-tuning of CLIP practical on bandwidth-limited clients instead of the millions of parameters required by full-model or adapter-based baselines.
  • On the 11 datasets, CacheFL beats PromptFL by an average of +9.6 points in the IID setting, +7.8 points under Dirichlet non-IID partitioning, and +7.1 points under extreme non-IID partitioning, while also exceeding zero-shot CLIP on every dataset.
  • The federated training stage carries the largest accuracy gain: on UCF101, accuracy rises from 57.9 percent with only the synthetic initialization to 77.6 percent after federated fine-tuning.
  • Under the stated assumptions, the federated cache update converges at rate $O(1/t)$, so most of the accuracy improvement appears within the first few communication rounds.
  • The transmitted cache-model updates resist a standard gradient-inversion attack, with reconstructed images resembling noise rather than the original training images.

Reading between the lines

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

  • Editorial inference: the cache model is effectively a learned nearest-prototype memory, so the same federated protocol could be applied to any frozen encoder whose features are meaningful under cosine or Euclidean similarity, not only CLIP.
  • Editorial inference: because only the cache matrix is exchanged, standard secure-aggregation or differential-privacy techniques would be much cheaper to add than for full-model federated learning, potentially giving stronger privacy guarantees than the gradient-inversion test alone.
  • Editorial inference: the choice of the text-to-image model is a testable bottleneck; replacing the generative model with a different one, or using a public feature bank from another domain, should change accuracy in a predictable way if the class-balanced synthetic prior is what carries the non-IID robustness.
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

4 major / 6 minor

Summary. The paper proposes CacheFL, a federated fine-tuning method for CLIP in which only a lightweight Tip-Adapter-style cache model is trained and communicated. The cache model is initialized from class-balanced synthetic images generated by DALL·E; clients update the visual-feature cache by SGD on a cross-entropy loss over fused CLIP and cache logits, and the server aggregates the cache updates with FedAvg. The paper reports accuracy improvements over FedAvg, FedProx, CLIP, AdapterFL, and PromptFL on 11 datasets under IID, Dirichlet, and extreme non-IID splits, reduced communication and computation costs, qualitative resistance to gradient inversion, and a convergence theorem claiming O(1/t) convergence.

Significance. If the empirical results are reproducible, the underlying idea is practically valuable: it is simple, communication-light, and improves on zero-shot CLIP in federated settings. The paper includes broad dataset coverage, ablations of synthetic initialization and of the number of synthetic images, and a resource-consumption comparison. However, the theoretical section overstates the guarantees: Theorem 1 relies on strong convexity that is not satisfied by the objective actually trained on clients. The empirical claims also need stronger support, in the form of error bars, validation-based hyperparameter selection, and a quantitative privacy evaluation, before they can be taken at face value.

major comments (4)
  1. [§IV-B, Assumption 2 and §IV-C, Theorem 1] The convergence proof uses Assumption 2 in Eq. (25) to obtain the contraction term (1−η_t μ)||c_k^t − c*||^2, and Theorem 1 inherits this strong-convexity requirement. However, the objective actually minimized on clients (Algorithm 1 lines 17–18; Eqs. (6)–(8)) has logits = f_vision W_text^T + α exp(−β(1−f_vision W_1)) W_2^T, and cross-entropy of these logits is generally non-convex in W_1 because of the exponential/RBF term. The manuscript asserts strong convexity without proof, saying only that Assumptions 1 and 2 are "canonical assumptions" for linear regression with softmax classification. Consequently, the O(1/t) bound in Eq. (60) is not established for CacheFL as implemented. The authors should either prove the required property, replace Theorem 1 with a nonconvex FL convergence analysis (e.g., convergence to stationarity under bounded-gradient assumptions), or explicitly restrict the theorem to a simplified convex surrogate and state that the experiments do not satisfy its assumptions.
  2. [§V-C, Table III] The hyperparameters α and β are selected as the values that maximize accuracy (α=0.5, β=1.0) and these values are then used in the headline comparisons of Table I. Table III does not state which dataset is used or whether a validation split was held out, so the tuning appears to be performed on the evaluation set. This can inflate the reported gains, especially because the sensitivity to α is sharp (accuracy drops from 64.1 to 51.4 as α goes from 0.5 to 2.0). Please use a validation split for hyperparameter selection, report which dataset Table III refers to, and ideally show sensitivity across all 11 datasets.
  3. [§V-A and §V-B, Table I] All accuracy numbers in Table I are reported without error bars, standard deviations, or the number of random seeds. In the small-data regime considered (16 images per class) with random non-IID partitioning, a single run per cell is insufficient to support claims such as "consistently achieves superior performance." Report means and standard deviations over at least three seeds and random partitions, at minimum for the averaged table and for the datasets with the largest variance.
  4. [§V-B, Privacy Preservation and Figure 7] The privacy evaluation is qualitative and compares a full-model FedAvg baseline against CacheFL's cache-model updates. Gradient inversion success depends strongly on model dimension and structure, so this comparison does not isolate the privacy protection of CacheFL. Provide quantitative reconstruction metrics (e.g., attack success rate, mean squared error, or learned perceptual similarity) and compare against baselines with comparable parameter counts, such as PromptFL or AdapterFL. Also clarify what the server observes about local data through the gradients of W_1 with respect to f_vision, since these gradients may carry information about local class statistics.
minor comments (6)
  1. [§IV-B, Assumption 2] Assumption 2 contains a typo: strong convexity should have (μ/2)||v−w||^2, not (L/2)||v−w||^2, since the proof and the rest of the paper use μ.
  2. [Table I and Figure 8] The dataset name is misspelled as "StandfordCars" in Table I(j) and in Figure 8; the correct spelling is "StanfordCars."
  3. [Figure 6] The labels in Figure 6 are garbled: "Ca cheFL" appears in panel (a), and the number "12712932" looks like an artifact. Please clean the figure labels and ensure the caption matches the reported values.
  4. [Figure 5] Several panels in Figure 5 contain stray numbers near the y-axis (e.g., "22", "28", "16", "38", "40") that appear to be residual axis labels or editing artifacts; please regenerate the figure.
  5. [§V-A, Communication Overhead and Figure 6a] The communication cost for CacheFL in Figure 6a is reported as 102,400 parameters per round, but the cache matrix W_1 has dimension C×N×K; for ImageNet (N=1000, K=16) this would be substantially larger than 102,400. Please specify which dataset and which parameter count are used so that the resource comparison is reproducible.
  6. [§V-C, Table II] The ablation "w.o. Training" is compared with PromptFL in the text, but the table shows only PromptFL and CacheFL rows; it would be clearer to include the full comparison or to state explicitly that PromptFL values are taken from Table I.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: CacheFL's cache model is inherited from external Tip-Adapter, its FL convergence proof is a standard analysis, and no prediction reduces to a fitted input or self-citation.

full rationale

The derivation chain is self-contained. The cache-model equations (5)-(7) are explicitly attributed to Tip-Adapter [15], an external baseline, rather than to the authors' own prior work. Federated aggregation follows FedAvg (Eq. 1), and the convergence proof in Theorem 1 is a standard federated-learning analysis under Assumptions 1-4. The reported accuracy gains are benchmarked against external baselines on held-out test sets. The listed weaknesses are correctness and experimental-design concerns, not circularity: Assumption 2 asserts mu-strong convexity for a loss that is generally non-convex in W1 due to the RBF term inside cross-entropy, and the hyperparameters alpha and beta are tuned on the evaluation set. However, the theorem does not assume the empirical conclusion it is used to support, and no fitted parameter is renamed as a prediction. No equation is defined in terms of the quantity it predicts, and no load-bearing claim depends on a self-citation. Therefore the paper shows no significant circularity.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The cache-model fusion comes from Tip-Adapter and the aggregation from FedAvg; the method-specific assumptions are the informativeness of DALL-E images for real classes and a shared label set. The convergence proof requires L-smoothness, strong convexity, and bounded-variance assumptions that are not verified for the actual RBF-based loss. Hyperparameters alpha, beta, K, and the learning rate are chosen by hand or by grid search, and the empirical learning-rate schedule does not match the theoretical one.

free parameters (4)
  • alpha (fusion weight) = 0.5
    Chosen by grid search in Table III because it maximized accuracy; used to blend CLIP logits with cache logits in Equation (7).
  • beta (cache logit scaling) = 1.0
    Chosen by grid search in Table III; controls the exponential distance kernel in Equation (6) and affects accuracy.
  • K (number of synthetic images per class) = 16
    Selected in Figure 8 as the largest tested value; it sets the cache size and the communication and computation overhead.
  • learning rate = 0.001 (constant)
    Fixed hyperparameter for SGD in the experiments; the convergence theorem instead uses the decreasing schedule eta_t = 2/(mu(t+gamma)), so the theoretical and empirical protocols do not match.
assumptions (5)
  • ad hoc to paper Local objectives are L-smooth and µ-strongly convex (Assumptions 1 and 2).
    Section IV-B; the actual loss in Eq. (7) includes the RBF term exp(-beta(1 - f_vision W1)), so cross-entropy in W1 is non-convex; no justification is provided.
  • domain assumption Bounded local gradient variance and bounded expected gradient norm (Assumptions 3 and 4).
    Section IV-B; standard in federated-learning analyses but not checked for the CacheFL cache model and the RBF loss.
  • domain assumption DALL-E synthetic images for each class are informative proxies for real images under CLIP features, so the cache initialization improves accuracy.
    Section III-A and ablation Table II; gains over random initialization depend on this, and the paper notes performance degrades when the synthetic distribution gap is large.
  • domain assumption All clients and the server share the same set of N class labels, so W2 one-hot encoding and text encoder class prompts are consistent.
    Section III-C and Algorithm 1; the server generates images for N categories and clients use the same 'Input N', so label heterogeneity across clients is not handled.
  • domain assumption CLIP's image and text encoders remain frozen during local training, so the gradient update is only over W1.
    Section III-C and Figure 4; the method treats CLIP as fixed, which is an assumption about feature stability rather than a derived property.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CacheFL: Privacy-Preserving and Efficient Federated Cache Model Fine-Tuning for Vision-Language Models." pith.science (2026). https://pith.science/paper/EYHBI37V

@misc{pith2026250505130,
  author       = {Pith},
  title        = {Pith review of: CacheFL: Privacy-Preserving and Efficient Federated Cache Model Fine-Tuning for Vision-Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EYHBI37V}},
  note         = {Machine review of arXiv:2505.05130}
}
read the original abstract

Large pre-trained Vision-Language Models (VLMs), such as Contrastive Language-Image Pre-training (CLIP), have exhibited remarkable zero-shot performance across various image classification tasks. Fine-tuning these models on domain-specific datasets further enhances their effectiveness for downstream applications. However, fine-tuning in cloud environments raises significant concerns regarding data security and privacy. Federated Learning (FL) offers a decentralized solution by enabling model training across local clients without centralizing sensitive data, but the high communication and computation costs of transmitting full pre-trained models during training limit its scalability. Additionally, non-Independent and Identically Distributed (non-IID) data across local clients can negatively impact model convergence and performance. To address these challenges, we propose CacheFL, a novel federated learning method that replaces traditional full model fine-tuning with lightweight cache model fine-tuning. The cache model is initialized using a class-balanced dataset generated by a generative pre-trained model, effectively mitigating the impact of non-IID data. This cache model is then distributed to local clients for fine-tuning, and the updated parameters from each client are aggregated on the server and redistributed. With the updated cache model, the classification performance of CLIP is improved after just a few epochs. By limiting the training and communication to the cache model, CacheFL significantly reduces resource demands while ensuring data privacy and security. Extensive experiments conducted on ImageNet and 10 additional datasets demonstrate that CacheFL outperforms traditional approaches in terms of classification accuracy, resource efficiency, and privacy preservation.

Figures

Figures reproduced from arXiv: 2505.05130 by the authors.

Figure 1
Figure 1. Performance comparison on the Extreme non-IID setting. Each axis [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The Structure of CLIP. Each backbone contains an image encoder [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Class-balanced Synthetic Dataset Generation. We adopt DALL·E to [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: The workflow of CacheFL Federated Training. Each client includes a cache model (with the visual feature weights being trainable and the one-hot [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Convergence Speed. The y-axis shows test accuracy (%), and the x-axis shows communication rounds. Across 4 datasets (ImageNet, Stanford Cars, [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Resource Consumption. We evaluate the communication cost based [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 8
Figure 8. Figure 8: Performance with the Number of Synthesized Images. DALL·E generates images for each category in increments of 1 to 16. [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 7 canonical work pages

  1. [1]

    Vision-language models for vision tasks: A survey,

    J. Zhang, J. Huang, S. Jin, and S. Lu, “Vision-language models for vision tasks: A survey,”IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    2024

  2. [2]

    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

    2021

  3. [3]

    Maple: Multi-modal prompt learning,

    M. U. Khattak, H. Rasheed, M. Maaz, S. Khan, and F. S. Khan, “Maple: Multi-modal prompt learning,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 19 113–19 122

    2023

  4. [4]

    Edge intelligence: Paving the last mile of artificial intelligence with edge computing,

    Z. Zhou, X. Chen, E. Li, L. Zeng, K. Luo, and J. Zhang, “Edge intelligence: Paving the last mile of artificial intelligence with edge computing,”Proceedings of the IEEE, vol. 107, no. 8, pp. 1738–1762, 2019

    2019

  5. [5]

    Advances and open problems in federated learning,

    P. Kairouz, H. B. McMahan, B. Avent, A. Bellet, M. Bennis, A. N. Bhagoji, K. Bonawitz, Z. Charles, G. Cormode, R. Cummingset al., “Advances and open problems in federated learning,”Foundations and Trends® in Machine Learning, vol. 14, no. 1–2, pp. 1–210, 2021

    2021

  6. [6]

    Adaptive asynchronous federated learning in resource-constrained edge computing,

    J. Liu, H. Xu, L. Wang, Y . Xu, C. Qian, J. Huang, and H. Huang, “Adaptive asynchronous federated learning in resource-constrained edge computing,”IEEE Transactions on Mobile Computing, vol. 22, no. 2, pp. 674–690, 2021

    work page 2021

  7. [7]

    Fedhome: Cloud-edge based personalized federated learning for in-home health monitoring,

    Q. Wu, X. Chen, Z. Zhou, and J. Zhang, “Fedhome: Cloud-edge based personalized federated learning for in-home health monitoring,”IEEE Transactions on Mobile Computing, vol. 21, no. 8, pp. 2818–2832, 2020

    2020

  8. [8]

    Communication-efficient federated learning,

    M. Chen, N. Shlezinger, H. V . Poor, Y . C. Eldar, and S. Cui, “Communication-efficient federated learning,”Proceedings of the Na- tional Academy of Sciences, vol. 118, no. 17, p. e2024789118, 2021

    2021

Show all 41 references
  1. [9]

    Heterogeneity-aware cooperative federated edge learning with adaptive computation and communication compression,

    Z. Zhang, Z. Gao, Y . Guoh, and Y . Gong, “Heterogeneity-aware cooperative federated edge learning with adaptive computation and communication compression,”IEEE Transactions on Mobile Computing, 2024

    2024

  2. [10]

    Com- putation and communication efficient federated learning with adaptive model pruning,

    Z. Jiang, Y . Xu, H. Xu, Z. Wang, J. Liu, Q. Chen, and C. Qiao, “Com- putation and communication efficient federated learning with adaptive model pruning,”IEEE Transactions on Mobile Computing, vol. 23, no. 3, pp. 2003–2021, 2023

    2003

  3. [11]

    Federated learning on non-iid data silos: An experimental study,

    Q. Li, Y . Diao, Q. Chen, and B. He, “Federated learning on non-iid data silos: An experimental study,” in2022 IEEE 38th international conference on data engineering (ICDE). IEEE, 2022, pp. 965–978

    2022

  4. [12]

    Fedsiam- da: Dual-aggregated federated learning via siamese network for non-iid data,

    X. Wang, Y . Wang, M. Yang, X. Wu, L. Fan, S. Heet al., “Fedsiam- da: Dual-aggregated federated learning via siamese network for non-iid data,”IEEE Transactions on Mobile Computing, 2024

    2024

  5. [13]

    Overcoming noisy labels and non-iid data in edge federated learning,

    Y . Xu, Y . Liao, L. Wang, H. Xu, Z. Jiang, and W. Zhang, “Overcoming noisy labels and non-iid data in edge federated learning,”IEEE Trans- actions on Mobile Computing, 2024

    2024

  6. [14]

    Zero-shot text-to-image generation,

    A. Ramesh, M. Pavlov, G. Goh, S. Gray, C. V oss, A. Radford, M. Chen, and I. Sutskever, “Zero-shot text-to-image generation,” inInternational conference on machine learning. Pmlr, 2021, pp. 8821–8831

    2021

  7. [15]

    Tip-adapter: Training-free clip-adapter for better vision-language modeling,

    R. Zhang, R. Fang, W. Zhang, P. Gao, K. Li, J. Dai, Y . Qiao, and H. Li, “Tip-adapter: Training-free clip-adapter for better vision-language modeling,”arXiv preprint arXiv:2111.03930, 2021

    2021 arXiv

  8. [16]

    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

    2022

  9. [17]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, “An image is worth 16x16 words: Transformers for image recognition at scale,”arXiv preprint arXiv:2010.11929, 2020

    2010 arXiv

  10. [18]

    Attention is all you need,

    A. Vaswani, “Attention is all you need,”Advances in Neural Information Processing Systems, 2017

    2017

  11. [19]

    Parameter-efficient fine-tuning of large- scale pre-trained language models,

    N. Ding, Y . Qin, G. Yang, F. Wei, Z. Yang, Y . Su, S. Hu, Y . Chen, C.-M. Chan, W. Chenet al., “Parameter-efficient fine-tuning of large- scale pre-trained language models,”Nature Machine Intelligence, vol. 5, no. 3, pp. 220–235, 2023

    2023

  12. [20]

    On the effectiveness of adapter-based tuning for pretrained language model adaptation,

    R. He, L. Liu, H. Ye, Q. Tan, B. Ding, L. Cheng, J.-W. Low, L. Bing, and L. Si, “On the effectiveness of adapter-based tuning for pretrained language model adaptation,”arXiv preprint arXiv:2106.03164, 2021

    2021 arXiv

  13. [21]

    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

    2021 arXiv

  14. [22]

    A review of applications in federated learning,

    L. Li, Y . Fan, M. Tse, and K.-Y . Lin, “A review of applications in federated learning,”Computers & Industrial Engineering, vol. 149, p. 106854, 2020

    2020

  15. [23]

    Communication-efficient learning of deep networks from decentralized data,

    B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from decentralized data,” inArtificial intelligence and statistics. PMLR, 2017, pp. 1273– 1282

    2017

  16. [24]

    Promptfl: Let federated participants cooperatively learn prompts instead of models-federated learning in age of foundation model,

    T. Guo, S. Guo, J. Wang, X. Tang, and W. Xu, “Promptfl: Let federated participants cooperatively learn prompts instead of models-federated learning in age of foundation model,”IEEE Transactions on Mobile Computing, 2023

    2023

  17. [25]

    Federated optimization in heterogeneous networks,

    T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V . Smith, “Federated optimization in heterogeneous networks,”Proceedings of Machine learning and systems, vol. 2, pp. 429–450, 2020

    2020

  18. [26]

    Generative adversarial networks,

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Bengio, “Generative adversarial networks,” Communications of the ACM, vol. 63, no. 11, pp. 139–144, 2020. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    2020

  19. [27]

    Catastrophic forgetting and mode collapse in gans,

    H. Thanh-Tung and T. Tran, “Catastrophic forgetting and mode collapse in gans,” in2020 international joint conference on neural networks (ijcnn). IEEE, 2020, pp. 1–10

    2020

  20. [28]

    Imagenet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “Imagenet: A large-scale hierarchical image database,” in2009 IEEE conference on computer vision and pattern recognition. Ieee, 2009, pp. 248–255

    2009

  21. [29]

    Learning generative visual models from few training examples: An incremental bayesian approach tested on 101 object categories,

    L. Fei-Fei, R. Fergus, and P. Perona, “Learning generative visual models from few training examples: An incremental bayesian approach tested on 101 object categories,” in2004 conference on computer vision and pattern recognition workshop. IEEE, 2004, pp. 178–178

    2004

  22. [30]

    Describing textures in the wild,

    M. Cimpoi, S. Maji, I. Kokkinos, S. Mohamed, and A. Vedaldi, “Describing textures in the wild,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2014, pp. 3606–3613

    2014

  23. [31]

    Eurosat: A novel dataset and deep learning benchmark for land use and land cover classi- fication,

    P. Helber, B. Bischke, A. Dengel, and D. Borth, “Eurosat: A novel dataset and deep learning benchmark for land use and land cover classi- fication,”IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 12, no. 7, pp. 2217–2226, 2019

    2019

  24. [32]

    Fine- grained visual classification of aircraft,

    S. Maji, E. Rahtu, J. Kannala, M. Blaschko, and A. Vedaldi, “Fine- grained visual classification of aircraft,”arXiv preprint arXiv:1306.5151, 2013

    2013 arXiv

  25. [33]

    Food-101–mining discriminative components with random forests,

    L. Bossard, M. Guillaumin, and L. Van Gool, “Food-101–mining discriminative components with random forests,” inComputer vision– ECCV 2014: 13th European conference, zurich, Switzerland, September 6-12, 2014, proceedings, part VI 13. Springer, 2014, pp. 446–461

    2014

  26. [34]

    Automated flower classification over a large number of classes,

    M.-E. Nilsback and A. Zisserman, “Automated flower classification over a large number of classes,” in2008 Sixth Indian conference on computer vision, graphics & image processing. IEEE, 2008, pp. 722–729

    2008

  27. [35]

    Cats and dogs,

    O. M. Parkhi, A. Vedaldi, A. Zisserman, and C. Jawahar, “Cats and dogs,” in2012 IEEE conference on computer vision and pattern recog- nition. IEEE, 2012, pp. 3498–3505

    2012

  28. [36]

    3d object representations for fine-grained categorization,

    J. Krause, M. Stark, J. Deng, and L. Fei-Fei, “3d object representations for fine-grained categorization,” inProceedings of the IEEE interna- tional conference on computer vision workshops, 2013, pp. 554–561

    2013

  29. [37]

    Sun database: Large-scale scene recognition from abbey to zoo,

    J. Xiao, J. Hays, K. A. Ehinger, A. Oliva, and A. Torralba, “Sun database: Large-scale scene recognition from abbey to zoo,” in2010 IEEE computer society conference on computer vision and pattern recognition. IEEE, 2010, pp. 3485–3492

    2010

  30. [38]

    Ucf101: A dataset of 101 human actions classes from videos in the wild,

    K. Soomro, “Ucf101: A dataset of 101 human actions classes from videos in the wild,”arXiv preprint arXiv:1212.0402, 2012

    2012 arXiv

  31. [39]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

    2016

  32. [40]

    Deep leakage from gradients,

    L. Zhu, Z. Liu, and S. Han, “Deep leakage from gradients,”Advances in neural information processing systems, vol. 32, 2019

    2019

  33. [41]

    Gpt-4 technical report,

    J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkatet al., “Gpt-4 technical report,”arXiv preprint arXiv:2303.08774, 2023

    2023 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.