pith. sign in

arxiv: 2605.00578 · v2 · pith:PLRYCHO4new · submitted 2026-05-01 · 💻 cs.CV

Federated Distillation for Whole Slide Image via Gaussian-Mixture Feature Alignment and Curriculum Integration

Luru Jing , Cong Cong , Yanyuan Chen , Yongzhi Cao This is my paper

Pith reviewed 2026-05-21 00:07 UTC · model grok-4.3

classification 💻 cs.CV
keywords federated learningwhole slide imagesGaussian mixturefeature alignmentknowledge distillationcurriculum learningdigital pathologymulti-institutional
0
0 comments X

The pith

FedHD establishes that local Gaussian-mixture feature alignment with one-to-one synthetic distillation and curriculum integration outperforms baselines in federated whole slide image classification.

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

The paper proposes FedHD as a federated learning method for whole slide images that avoids exchanging model parameters. Each client aligns its features to a Gaussian mixture model and distills one synthetic feature representation for every real slide to maintain diagnostic variety. A curriculum schedule adds cross-site synthetic features to local training only after the model stops improving. This process supports different model architectures at each institution while keeping data private. Results on TCGA-IDH, CAMELYON16, and CAMELYON17 show consistent gains over existing federated and distillation approaches.

Core claim

By performing local Gaussian-mixture feature alignment to produce semantically rich synthetic features, applying one-to-one distillation to avoid compression loss, and progressively integrating cross-site synthetics via curriculum once local performance plateaus, the framework delivers higher accuracy in multi-institutional whole slide image tasks without sharing raw data or model weights.

What carries the argument

Local Gaussian-mixture feature alignment that produces one synthetic feature counterpart per real slide for subsequent one-to-one distillation and curriculum integration.

If this is right

  • Accuracy rises over state-of-the-art federated and distillation baselines on TCGA-IDH, CAMELYON16, and CAMELYON17.
  • Training remains compatible with varied multiple-instance learning architectures at different sites.
  • Only synthetic features are exchanged, keeping raw patient slides and model parameters private.
  • An optional module can reconstruct pseudo-patches from the synthetic embeddings to support interpretation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same alignment-plus-curriculum pattern could apply to other heterogeneous medical imaging tasks where sites cannot share raw scans.
  • If the synthetic features retain diagnostic signals across more than three sites, the approach may scale to larger federated networks.
  • Adding noise to the synthetic features before sharing could be tested as a way to strengthen privacy guarantees.
  • Comparing the method against direct feature averaging without curriculum would isolate the benefit of the staged integration schedule.

Load-bearing premise

Generating one synthetic counterpart per real slide via Gaussian-mixture alignment preserves enough diagnostic diversity that curriculum integration improves local performance without adding distribution shift or bias.

What would settle it

Measure whether local validation accuracy rises or falls after the curriculum phase begins adding cross-site synthetic features; a consistent drop would indicate the integration step fails to help.

Figures

Figures reproduced from arXiv: 2605.00578 by Cong Cong, Luru Jing, Yanyuan Chen, Yongzhi Cao.

Figure 1
Figure 1. Figure 1: Overview of the FedHD Framework. ① Each institution c distills its local WSIs into a set of synthetic slides ({h c i } N i=1) through the local Gaussian-mixture feature distillation process. {h c i } N i=1 are then uploaded to a central server, which aggregates them and constructs a global synthetic dataset H (c) global for each client by excluding that client’s own data. ② Each institution subsequently tr… view at source ↗
Figure 2
Figure 2. Figure 2: t-SNE visualization of patch-level feature embeddings from real slides and various ablated versions of FedHD. CAMELYON16 is used for this demonstration as it provides patch-level tumor annotations. Each point represents a patch embedding, color-coded by class (Normal vs. Tumor). We additionally report the corresponding local model performance when trained using real slides or synthetic samples to facilitat… view at source ↗
Figure 3
Figure 3. Figure 3: Baseline synthetic images are unrealistic. The PPR mod￾ule enables realistic on-demand reconstructions without adding training overhead. The first row shows normal patches and the second row shows tumor patches. gles to capture the full complexity of real slide features, limiting its standalone effectiveness. Impact of O2O: As shown in [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of heatmaps from a model trained with naive data concatenation versus CBF. CBF not only produces more precise and diagnostically relevant regions in successful cases (first row) but also corrects predictions by better localizing pathological cues in cases where the naive approach fails (second row) [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Classification performances with different number of synthetic patches per slide (T) (upper) and different number of Gaussian mixture components (M) (lower). B.1. Ablation Study on Different Number of Synthetic Patches per Slide (T). We analyze how the number of synthetic patches per slide (T) affects model performance by varying it from 50 to 5000. As shown in [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Classification performances with different curriculum threshold (t0) (upper) and different noise robustness parameter in GCE loss (q) (lower). B.3. Ablation Study on Different Curriculum Threshold (t0). We investigate the effect of the curriculum threshold t0, which determines when synthetic data from other clients are introduced into local training. As shown in [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: More visualizations of synthetic patches using Pseudo-Patch Reconstruction. show that pathology guidance leads to more structured, diverse, and morphologically realistic patches that better resemble true histological patterns. E. Evaluation using Distilled Data. To assess the quality of distilled samples, we train a shared local MIL model (CLAM) using only the synthetic data generated by different FL+DD me… view at source ↗
read the original abstract

Federated learning (FL) offers a promising framework for collaborative digital pathology by enabling model training across institutions. However, real-world deployments face heterogeneity arising from diverse multiple instance learning (MIL) architectures and heterogeneous feature extractors across institutions. We propose FedHD, a novel FL framework that performs local Gaussian-mixture feature alignment tailored for WSI analysis. Instead of exchanging model parameters, each client independently distills semantically rich synthetic feature representations aligned with the distribution of real WSIs. To preserve diagnostic diversity, FedHD adopts a one-to-one distillation strategy, generating a synthetic counterpart for each real slide to avoid over-compression. During federation, a curriculum-based integration strategy progressively incorporates cross-site synthetic features into local training once performance plateaus. Furthermore, an optional interpretation module reconstructs pseudo-patches from synthetic embeddings, enhancing transparency. FedHD is architecture-agnostic, privacy-preserving, and supports personalized yet collaborative training across diverse institutions. Experiments on TCGA-IDH, CAMELYON16, and CAMELYON17 show that FedHD consistently outperforms state-of-the-art federated and distillation baselines.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The paper introduces FedHD, a federated learning framework for whole slide image (WSI) analysis in digital pathology. It performs local Gaussian-mixture feature alignment to generate synthetic feature representations, uses one-to-one distillation to preserve diagnostic diversity, and employs a curriculum-based integration strategy to incorporate cross-site synthetic features into local training once performance plateaus. The method is claimed to be architecture-agnostic and privacy-preserving. Experiments on TCGA-IDH, CAMELYON16, and CAMELYON17 datasets demonstrate consistent outperformance against state-of-the-art federated and distillation baselines.

Significance. If the empirical results hold under rigorous validation, FedHD could advance collaborative model training across institutions without sharing sensitive patient data or model parameters, addressing key challenges of heterogeneity in MIL architectures and feature extractors in computational pathology. The use of synthetic features and curriculum learning offers a novel approach to knowledge transfer in federated settings.

major comments (1)
  1. Curriculum integration strategy: The central claim depends on progressively incorporating cross-site synthetic features once local performance plateaus. With heterogeneous MIL architectures and feature extractors across clients, plateaus can arise from local overfitting rather than global readiness; the manuscript must demonstrate that plateau detection is robust to site heterogeneity and does not inject distribution shift or bias upon integration. Provide ablations on integration timing and synchronization across clients.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address the major comment point by point below and indicate where revisions will be made to strengthen the work.

read point-by-point responses

  1. Referee: Curriculum integration strategy: The central claim depends on progressively incorporating cross-site synthetic features once local performance plateaus. With heterogeneous MIL architectures and feature extractors across clients, plateaus can arise from local overfitting rather than global readiness; the manuscript must demonstrate that plateau detection is robust to site heterogeneity and does not inject distribution shift or bias upon integration. Provide ablations on integration timing and synchronization across clients.

    Authors: We agree that demonstrating the robustness of plateau detection under site heterogeneity is important to support the central claim. In FedHD, each client independently monitors its local validation performance and triggers integration once a plateau is reached, allowing sites to stabilize before cross-site synthetic features are incorporated. To address the referee's concern, we will add ablations in the revised manuscript that vary integration timing (e.g., early vs. late plateau detection) and test synchronization across clients using different MIL architectures and feature extractors on CAMELYON17. These experiments will include metrics on performance and feature distribution similarity (such as MMD) before and after integration to verify that no significant distribution shift or bias is introduced. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper describes FedHD as an empirical federated learning method that applies local Gaussian-mixture feature alignment for WSI, performs one-to-one distillation to generate synthetic counterparts per real slide, and uses curriculum integration of cross-site synthetics once local performance plateaus. These steps are presented as design choices evaluated through experiments on public datasets (TCGA-IDH, CAMELYON16, CAMELYON17) showing outperformance over baselines. No equations, self-definitional reductions, fitted parameters renamed as predictions, or load-bearing self-citations appear in the provided text. The central claims rest on independent empirical validation rather than any derivation that collapses to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Review is based solely on the abstract; specific free parameters, axioms, and entities cannot be exhaustively identified without the full text. The method appears to rest on standard federated learning and feature modeling assumptions.

axioms (2)
  • domain assumption Synthetic features distilled locally can substitute for model parameter exchange while preserving utility across heterogeneous clients
    Core premise of the distillation strategy described in the abstract
  • domain assumption Curriculum integration of cross-site synthetics after local performance plateaus improves overall training without negative transfer
    Key mechanism for federation phase in the proposed framework

pith-pipeline@v0.9.0 · 5728 in / 1514 out tokens · 52368 ms · 2026-05-21T00:07:08.805283+00:00 · methodology

Review history (2 revisions) →

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

87 extracted references · 87 canonical work pages · 1 internal anchor

  1. [1]

    Nature , volume=

    A pathology foundation model for cancer diagnosis and prognosis prediction , author=. Nature , volume=. 2024 , publisher=

    work page 2024

  2. [2]

    Nature biomedical engineering , volume=

    Data-efficient and weakly supervised computational pathology on whole-slide images , author=. Nature biomedical engineering , volume=. 2021 , publisher=

    work page 2021

  3. [3]

    Advances in neural information processing systems , volume=

    Transmil: Transformer based correlated multiple instance learning for whole slide image classification , author=. Advances in neural information processing systems , volume=

    work page

  4. [4]

    Nature Medicine , publisher=

    Towards a General-Purpose Foundation Model for Computational Pathology , author=. Nature Medicine , publisher=

    work page

  5. [5]

    arXiv preprint arXiv:2409.09173 , year=

    Phikon-v2, a large and public feature extractor for biomarker prediction , author=. arXiv preprint arXiv:2409.09173 , year=

    work page arXiv

  6. [6]

    2021 17th International Conference on Mobility, Sensing and Networking (MSN) , pages=

    Fedhe: Heterogeneous models and communication-efficient federated learning , author=. 2021 17th International Conference on Mobility, Sensing and Networking (MSN) , pages=. 2021 , organization=

    work page 2021

  7. [7]

    Proceedings of Machine learning and systems , volume=

    Federated optimization in heterogeneous networks , author=. Proceedings of Machine learning and systems , volume=

    work page

  8. [8]

    IEEE transactions on neural networks and learning systems , volume=

    Towards personalized federated learning , author=. IEEE transactions on neural networks and learning systems , volume=. 2022 , publisher=

    work page 2022

  9. [9]

    arXiv preprint arXiv:2106.06042 , year=

    Fedbabu: Towards enhanced representation for federated image classification , author=. arXiv preprint arXiv:2106.06042 , year=

    work page arXiv

  10. [10]

    Advances in Neural Information Processing Systems , volume=

    Parameterized knowledge transfer for personalized federated learning , author=. Advances in Neural Information Processing Systems , volume=

    work page

  11. [11]

    IEEE Transactions on Medical Imaging , volume =

    From detection of individual metastases to classification of lymph node status at the patient level: the. IEEE Transactions on Medical Imaging , volume =. 2019 , pages =

    work page 2019

  12. [12]

    JAMA , volume=

    Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer , author=. JAMA , volume=. 2017 , publisher=

    work page 2017

  13. [13]

    Isocitrate dehydrogenase

    Liu, Sidong and Shah, Zubair and Sav, Aydin and Russo, Carlo and Berkovsky, Shlomo and Qian, Yi and Coiera, Enrico and Di Ieva, Antonio , journal=. Isocitrate dehydrogenase

    work page

  14. [14]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , year=

    Morphological Prototyping for Unsupervised Slide Representation Learning in Computational Pathology , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , year=

    work page

  15. [15]

    Advances in neural information processing systems , volume=

    Generalized cross entropy loss for training deep neural networks with noisy labels , author=. Advances in neural information processing systems , volume=

    work page

  16. [16]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Dual-stream multiple instance learning network for whole slide image classification with self-supervised contrastive learning , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    work page

  17. [17]

    Nature medicine , volume=

    Federated learning for predicting clinical outcomes in patients with COVID-19 , author=. Nature medicine , volume=. 2021 , publisher=

    work page 2021

  18. [18]

    arXiv preprint arXiv:2203.16622 , year=

    Federated learning for the classification of tumor infiltrating lymphocytes , author=. arXiv preprint arXiv:2203.16622 , year=

    work page arXiv

  19. [19]

    Medical image analysis , volume=

    Federated learning for computational pathology on gigapixel whole slide images , author=. Medical image analysis , volume=. 2022 , publisher=

    work page 2022

  20. [20]

    Nature communications , volume=

    Mining multi-center heterogeneous medical data with distributed synthetic learning , author=. Nature communications , volume=. 2023 , publisher=

    work page 2023

  21. [21]

    Nature Communications , volume=

    Privacy risks of whole-slide image sharing in digital pathology , author=. Nature Communications , volume=. 2023 , publisher=

    work page 2023

  22. [22]

    International Journal of Machine Learning and Cybernetics , volume=

    A survey on federated learning: challenges and applications , author=. International Journal of Machine Learning and Cybernetics , volume=. 2023 , publisher=

    work page 2023

  23. [23]

    Scientific Reports , volume=

    A hybrid learning network with progressive resizing and PCA for diagnosis of cervical cancer on WSI slides , author=. Scientific Reports , volume=. 2025 , publisher=

    work page 2025

  24. [24]

    Pattern recognition , volume=

    Federated learning for medical image analysis: A survey , author=. Pattern recognition , volume=. 2024 , publisher=

    work page 2024

  25. [25]

    Frontiers in Medicine , volume=

    Abnormality-aware multimodal learning for WSI classification , author=. Frontiers in Medicine , volume=. 2025 , publisher=

    work page 2025

  26. [26]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Federated Learning with Domain Shift Eraser , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    work page

  27. [27]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    A New Federated Learning Framework Against Gradient Inversion Attacks , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

  28. [28]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    HistoFS: Non-IID Histopathologic Whole Slide Image Classification via Federated Style Transfer with RoI-Preserving , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    work page

  29. [29]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Focus: Knowledge-enhanced adaptive visual compression for few-shot whole slide image classification , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    work page

  30. [30]

    Proceedings of the 2024 7th International Conference on Machine Vision and Applications , pages=

    Survival prediction across diverse cancer types using neural networks , author=. Proceedings of the 2024 7th International Conference on Machine Vision and Applications , pages=

    work page 2024

  31. [31]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Promptable representation distribution learning and data augmentation for gigapixel histopathology WSI analysis , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

  32. [32]

    International Conference on Medical Image Computing and Computer-Assisted Intervention , pages=

    Pathm3: A multimodal multi-task multiple instance learning framework for whole slide image classification and captioning , author=. International Conference on Medical Image Computing and Computer-Assisted Intervention , pages=. 2024 , organization=

    work page 2024

  33. [33]

    Poligon: A System for Parallel Problem Solving

    Rice, James. Poligon: A System for Parallel Problem Solving

    work page

  34. [34]

    Transfer of Rule-Based Expertise through a Tutorial Dialogue

    Clancey, William J. Transfer of Rule-Based Expertise through a Tutorial Dialogue

    work page

  35. [35]

    The Engineering of Qualitative Models

    Clancey, William J. The Engineering of Qualitative Models

    work page

  36. [36]

    2017 , eprint=

    Attention Is All You Need , author=. 2017 , eprint=

    work page 2017

  37. [37]

    Pluto: The 'Other' Red Planet

    NASA. Pluto: The 'Other' Red Planet

    work page

  38. [38]

    Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

    Dataset condensation with distribution matching , author=. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

    work page

  39. [39]

    arXiv preprint arXiv:2101.05428 , year=

    Federated learning: Opportunities and challenges , author=. arXiv preprint arXiv:2101.05428 , year=

    work page arXiv

  40. [40]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Point transformer with federated learning for predicting breast cancer her2 status from hematoxylin and eosin-stained whole slide images , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    work page

  41. [41]

    Nature medicine , volume=

    Federated learning for predicting histological response to neoadjuvant chemotherapy in triple-negative breast cancer , author=. Nature medicine , volume=. 2023 , publisher=

    work page 2023

  42. [42]

    Siloed federated learning for multi-centric histopathology datasets , author=. Domain Adaptation and Representation Transfer, and Distributed and Collaborative Learning: Second MICCAI Workshop, DART 2020, and First MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Lima, Peru, October 4--8, 2020, Proceedings 2 , pages=. 2020 , organization=

    work page 2020

  43. [43]

    Future Generation Computer Systems , volume=

    KDRSFL: A knowledge distillation resistance transfer framework for defending model inversion attacks in split federated learning , author=. Future Generation Computer Systems , volume=. 2025 , publisher=

    work page 2025

  44. [44]

    Sok: On gradient leakage in federated learning,

    SoK: On Gradient Leakage in Federated Learning , author=. arXiv preprint arXiv:2404.05403 , year=

    work page arXiv

  45. [45]

    Dataset Distillation

    Tongzhou Wang and Jun. Dataset Distillation , journal =. 2018 , url =. 1811.10959 , timestamp =

    work page internal anchor Pith review arXiv 2018

  46. [46]

    Reinforcement Learning-based Decentralized Optimal Control for Large- Scale Multi-agent System by Using Neural Networks and Discrete-time Mean Field Games

    Sucholutsky, Ilia and Schonlau, Matthias , year=. Soft-Label Dataset Distillation and Text Dataset Distillation , url=. doi:10.1109/ijcnn52387.2021.9533769 , booktitle=

    work page doi:10.1109/ijcnn52387.2021.9533769 2021

  47. [47]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    FedMut: Generalized federated learning via stochastic mutation , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

    work page

  48. [48]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Fedproto: Federated prototype learning across heterogeneous clients , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

    work page

  49. [49]

    2009 IEEE conference on computer vision and pattern recognition , pages=

    Imagenet: A large-scale hierarchical image database , author=. 2009 IEEE conference on computer vision and pattern recognition , pages=. 2009 , organization=

    work page 2009

  50. [50]

    European Conference on Computer Vision , pages=

    Unlocking the potential of federated learning: The symphony of dataset distillation via deep generative latents , author=. European Conference on Computer Vision , pages=. 2024 , organization=

    work page 2024

  51. [51]

    , author=

    Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis. , author=. iclr , year=

    work page

  52. [52]

    arXiv preprint arXiv:2408.09709 , year=

    Dataset distillation for histopathology image classification , author=. arXiv preprint arXiv:2408.09709 , year=

    work page arXiv

  53. [53]

    and Coiera, Enrico and Liu, Sidong , title =

    Cong, Cong and Song, Yang and Di Ieva, Antonio and Chou, Angela and Gill, Anthony J. and Coiera, Enrico and Liu, Sidong , title =. npj Digital Medicine , year =. doi:10.1038/s41746-026-02710-6 , url =

    work page doi:10.1038/s41746-026-02710-6

  54. [54]

    International Conference on Medical Image Computing and Computer-Assisted Intervention , pages=

    Communication-efficient federated skin lesion classification with generalizable dataset distillation , author=. International Conference on Medical Image Computing and Computer-Assisted Intervention , pages=. 2023 , organization=

    work page 2023

  55. [55]

    2023 International Joint Conference on Neural Networks (IJCNN) , pages=

    Federated learning via decentralized dataset distillation in resource-constrained edge environments , author=. 2023 International Joint Conference on Neural Networks (IJCNN) , pages=. 2023 , organization=

    work page 2023

  56. [56]

    2022 , eprint=

    Dataset Distillation for Medical Dataset Sharing , author=. 2022 , eprint=

    work page 2022

  57. [57]

    2024 , eprint=

    Progressive trajectory matching for medical dataset distillation , author=. 2024 , eprint=

    work page 2024

  58. [58]

    2023 , eprint=

    Federated Learning via Decentralized Dataset Distillation in Resource-Constrained Edge Environments , author=. 2023 , eprint=

    work page 2023

  59. [59]

    2025 , eprint=

    Secure Federated Data Distillation , author=. 2025 , eprint=

    work page 2025

  60. [60]

    2021 , eprint=

    Distilled One-Shot Federated Learning , author=. 2021 , eprint=

    work page 2021

  61. [61]

    2024 , eprint=

    Dataset Distillation-based Hybrid Federated Learning on Non-IID Data , author=. 2024 , eprint=

    work page 2024

  62. [62]

    International conference on machine learning , pages=

    Agnostic federated learning , author=. International conference on machine learning , pages=. 2019 , organization=

    work page 2019

  63. [63]

    2020 fourth world conference on smart trends in systems, security and sustainability (WorldS4) , pages=

    Survey of personalization techniques for federated learning , author=. 2020 fourth world conference on smart trends in systems, security and sustainability (WorldS4) , pages=. 2020 , organization=

    work page 2020

  64. [64]

    Journal of Machine Learning Research , volume=

    Pfllib: A beginner-friendly and comprehensive personalized federated learning library and benchmark , author=. Journal of Machine Learning Research , volume=

    work page

  65. [65]

    Information Fusion , volume=

    FairDPFL-SCS: Fair Dynamic Personalized Federated Learning with strategic client selection for improved accuracy and fairness , author=. Information Fusion , volume=. 2025 , publisher=

    work page 2025

  66. [66]

    2021 , eprint=

    Model-Contrastive Federated Learning , author=. 2021 , eprint=

    work page 2021

  67. [67]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Model-contrastive federated learning , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    work page

  68. [68]

    arXiv preprint arXiv:2111.04263 , year=

    Federated learning based on dynamic regularization , author=. arXiv preprint arXiv:2111.04263 , year=

    work page arXiv

  69. [69]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Feddm: Iterative distribution matching for communication-efficient federated learning , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    work page

  70. [70]

    Proceedings of the AAAI Conference on Artificial Intelligence , author=

    FedMut: Generalized Federated Learning via Stochastic Mutation , volume=. Proceedings of the AAAI Conference on Artificial Intelligence , author=. 2024 , month=. doi:10.1609/aaai.v38i11.29146 , abstractNote=

    work page doi:10.1609/aaai.v38i11.29146 2024

  71. [71]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Image-to-image translation with conditional adversarial networks , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    work page

  72. [72]

    2022 , eprint=

    FedProto: Federated Prototype Learning across Heterogeneous Clients , author=. 2022 , eprint=

    work page 2022

  73. [73]

    European conference on computer vision , pages=

    Attention-challenging multiple instance learning for whole slide image classification , author=. European conference on computer vision , pages=. 2024 , organization=

    work page 2024

  74. [74]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Re-thinking model inversion attacks against deep neural networks , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    work page

  75. [75]

    Artificial intelligence and statistics , pages=

    Communication-efficient learning of deep networks from decentralized data , author=. Artificial intelligence and statistics , pages=. 2017 , organization=

    work page 2017

  76. [76]

    2022 IEEE symposium on security and privacy (SP) , pages=

    Membership inference attacks from first principles , author=. 2022 IEEE symposium on security and privacy (SP) , pages=. 2022 , organization=

    work page 2022

  77. [77]

    IEEE transactions on signal processing , volume=

    Gaussian mixture modeling by exploiting the Mahalanobis distance , author=. IEEE transactions on signal processing , volume=. 2008 , publisher=

    work page 2008

  78. [78]

    Proceedings of the 41st International Conference on Machine Learning , year =

    Overcoming Data and Model heterogeneities in Decentralized Federated Learning via Synthetic Anchors , author =. Proceedings of the 41st International Conference on Machine Learning , year =

    work page

  79. [79]

    Proceedings of the AAAI Conference on Artificial Intelligence (AAAI) , year=

    Towards Adversarially Robust Dataset Distillation by Curvature Regularization , author=. Proceedings of the AAAI Conference on Artificial Intelligence (AAAI) , year=

    work page

  80. [80]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops , month =

    Wei, Wei and De Schepper, Tom and Mets, Kevin , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops , month =. 2024 , pages =

    work page 2024

Showing first 80 references.