pith. sign in

arxiv: 2606.10250 · v1 · pith:MAU2RNKBnew · submitted 2026-06-08 · 💻 cs.LG · cs.AI

Multi-Level Analyzation of Imbalance to Resolve Non-IID-Ness in Federated Learning

Haengbok Chung , Jae Sung Lee This is my paper

Pith reviewed 2026-06-27 16:49 UTC · model grok-4.3

classification 💻 cs.LG cs.AI
keywords federated learningnon-IID dataclass imbalancedata heterogeneityloss functionmodel aggregationmedical imaging
0
0 comments X

The pith

FedBB corrects imbalance at case, class and client levels to reduce non-IID effects in federated learning.

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

The paper identifies three distinct imbalance levels that degrade performance when data across clients is non-identically distributed: imbalance inside each class, imbalance between classes, and differing skewness across clients. It introduces FedBB, which applies a Positive Negative Balanced loss during each client's local training to raise the weight of minority cases and classes, then uses Client Balanced Reweighting at the server to give more influence to models from less skewed clients. Experiments on X-ray and natural-image datasets show higher accuracy and faster convergence than prior federated methods while exchanging only limited statistics. The resulting global model is intended to classify every class correctly rather than favoring majority distributions.

Core claim

Defining imbalance at the inter-case, inter-class and inter-client levels allows a combination of PNB loss for local training and CBR reweighting for aggregation to produce a global model that accurately classifies all classes on heterogeneous client data.

What carries the argument

Three-level imbalance analysis together with the Positive Negative Balanced loss (which weights minority cases and classes) and Client Balanced Reweighting (which adjusts client influence during aggregation).

If this is right

  • Local training becomes more robust to highly skewed client datasets.
  • Global aggregation favors models from clients whose data is less skewed.
  • Only summary statistics are needed, limiting exposure of raw client data.
  • The same components can serve as a baseline for both generic and personalized federated learning.

Where Pith is reading between the lines

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

  • The three-level framing may apply to other modalities such as text or sensor streams that exhibit similar case-class-client skews.
  • Reducing reliance on full data sharing could support stricter privacy regulations in medical or financial federated settings.
  • A follow-up test could measure whether the method still helps when label noise or feature shift, rather than count imbalance, is the dominant non-IID source.

Load-bearing premise

The three imbalance levels are the main causes of non-IID degradation and PNB loss plus CBR reweighting can be applied without introducing fresh biases on actual client data distributions.

What would settle it

A controlled experiment on client data whose imbalance patterns fall outside the three defined levels, in which FedBB shows no accuracy gain over standard FedAvg.

Figures

Figures reproduced from arXiv: 2606.10250 by Haengbok Chung, Jae Sung Lee.

Figure 1
Figure 1. Figure 1: Illustration of various types of labels of multi-label classification with three categories. [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FedBB process comprises CBR and PNB loss functions. The weights used in the [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Blue bars denote the positive rate of the overall dataset and red bars denote the [PITH_FULL_IMAGE:figures/full_fig_p022_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Proportions of clients’ models in the model aggregation stage. NIH CXR 14, CIFAR [PITH_FULL_IMAGE:figures/full_fig_p022_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Test AUC plots of NIH CXR14 and CheXpert and the test accuracy plots of CIFAR [PITH_FULL_IMAGE:figures/full_fig_p024_5.png] view at source ↗
read the original abstract

Class imbalance is a common problem in deep learning that severely degrades performance. In federated learning (FL), it is a critical factor contributing to non-identically distributed data (non-IID). Building on several previous attempts, we define and analyze imbalance issues in FL at three levels: inter-case, inter-class, and inter-client. Inter-case imbalance addresses the imbalance in every single class; inter-class imbalance compares the number of data between different classes. Inter-client imbalance represents different skewness of local data between clients. Based on these concepts, we propose FedBB, which consists of two main components: (1) Positive Negative Balanced (PNB) loss function addresses the inter-case and inter-class imbalances in local training, enhancing generalization on highly skewed local client datasets. It optimizes both multi-label and multi-class classifications by assigning higher weights to minority cases or classes. (2) Client Balanced Reweighting (CBR) reweights clients based on inter-client imbalance during model aggregation, giving greater weight to models trained on less skewed datasets. Various experiments on X-ray and natural image datasets demonstrate that FedBB outperforms other algorithms in both performance and efficiency. Additionally, it requires limited statistical information, which is beneficial for privacy protection. Through ablation studies, we proved that PNB loss and CBR independently contribute to performance. As FedBB aims to build a global model that accurately classifies all classes, it can serve as a baseline for the generic and personalized FL.

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

2 major / 2 minor

Summary. The manuscript defines imbalance in federated learning at three levels (inter-case, inter-class, inter-client) as drivers of non-IID degradation and proposes FedBB, which combines a Positive Negative Balanced (PNB) loss for local training (reweighting minority cases/classes) with Client Balanced Reweighting (CBR) during aggregation (weighting clients by skewness). It claims that experiments on X-ray and natural-image datasets show FedBB outperforming baselines in accuracy and efficiency, that ablation studies demonstrate independent contributions from PNB and CBR, and that the method requires only limited statistical information, aiding privacy. The approach is positioned as a baseline for generic and personalized FL.

Significance. If the empirical claims are substantiated with full metrics and robustness checks, the multi-level imbalance taxonomy offers a structured lens on non-IID sources and the two-component method supplies a lightweight, privacy-preserving mitigation strategy that could be adopted as a reference implementation. The limited-statistics requirement is a concrete practical strength.

major comments (2)
  1. [Abstract] Abstract and experimental claims: the assertion that 'various experiments demonstrate that FedBB outperforms other algorithms in both performance and efficiency' and that 'ablation studies proved that PNB loss and CBR independently contribute' supplies no quantitative metrics, baseline names, dataset sizes, statistical tests, or cross-validation details. This directly undermines verification of the central outperformance claim.
  2. [Method / FedBB components] Proposal of PNB+CBR (implicitly in the method section): the claim that these components address the three defined imbalance levels without introducing new biases rests on the untested assumption that count-based skewness fully captures client heterogeneity. No stress tests are described on distributions containing feature skew, label noise, or client-size variation outside pure count statistics, which are common in real FL and could amplify variance under reweighting.
minor comments (2)
  1. [Title] Title uses 'Analyzation'; standard term is 'Analysis'.
  2. [PNB loss description] The statement that PNB 'optimizes both multi-label and multi-class classifications' is stated without clarifying whether the same weighting formula applies to both or whether separate formulations are used.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the opportunity to respond to the referee's comments. We provide point-by-point responses to the major comments and indicate planned revisions to the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract and experimental claims: the assertion that 'various experiments demonstrate that FedBB outperforms other algorithms in both performance and efficiency' and that 'ablation studies proved that PNB loss and CBR independently contribute' supplies no quantitative metrics, baseline names, dataset sizes, statistical tests, or cross-validation details. This directly undermines verification of the central outperformance claim.

    Authors: We acknowledge that the abstract lacks specific quantitative details. The full paper includes these in the experiments section (Section 4), with comparisons to baselines such as FedAvg and others on datasets like Chest X-ray and natural image datasets, showing outperformance in accuracy and efficiency. To address this, we will revise the abstract to include key metrics from our results, specific baseline names, dataset sizes, and mention of the evaluation methodology. This will make the claims more verifiable directly from the abstract. revision: yes

  2. Referee: [Method / FedBB components] Proposal of PNB+CBR (implicitly in the method section): the claim that these components address the three defined imbalance levels without introducing new biases rests on the untested assumption that count-based skewness fully captures client heterogeneity. No stress tests are described on distributions containing feature skew, label noise, or client-size variation outside pure count statistics, which are common in real FL and could amplify variance under reweighting.

    Authors: Our definitions and method focus explicitly on the three levels of count-based imbalance (inter-case, inter-class, inter-client) as introduced in the paper. The PNB loss and CBR are designed to mitigate these without assuming they capture all possible heterogeneity. We agree that additional heterogeneities like feature skew or label noise are not tested in our experiments, which are limited to the defined imbalances on the X-ray and natural image datasets. We will revise the manuscript to include a limitations subsection discussing these aspects and the potential for variance amplification, while maintaining that within the paper's scope, no new biases are introduced as reweighting uses only aggregate counts. revision: partial

Circularity Check

0 steps flagged

No circularity: empirical proposal with external dataset validation and no self-referential derivations

full rationale

The paper defines three imbalance levels (inter-case, inter-class, inter-client) as conceptual categories and proposes PNB loss plus CBR reweighting as an empirical method. It reports performance on X-ray and natural image datasets plus ablation studies showing independent contributions. No equations, fitted parameters, or predictions are presented that reduce by construction to quantities defined from the same data or prior self-citations. The central claims rest on experimental outcomes rather than any load-bearing self-citation chain or definitional equivalence. This is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No explicit free parameters, mathematical axioms, or newly postulated entities are described in the abstract; the contribution rests on standard federated averaging assumptions plus the empirical claim that the two proposed components mitigate the defined imbalances.

pith-pipeline@v0.9.1-grok · 5794 in / 1172 out tokens · 23298 ms · 2026-06-27T16:49:32.976339+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

68 extracted references · 15 canonical work pages · 4 internal anchors

  1. [1]

    Communication-efficient learning of deep networks from decentralized data

    Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Aguera y Ar- cas. Communication-efficient learning of deep networks from decentralized data. InArtifi- cial intelligence and statistics, pages 1273–1282. PMLR, 2017

    2017

  2. [2]

    The future of digital health with federated learning.NPJ digital medicine, 3(1):119, 2020

    Nicola Rieke, Jonny Hancox, Wenqi Li, Fausto Milletari, Holger R Roth, Shadi Albarqouni, Spyridon Bakas, Mathieu N Galtier, Bennett A Landman, Klaus Maier-Hein, et al. The future of digital health with federated learning.NPJ digital medicine, 3(1):119, 2020

    2020

  3. [3]

    Secure, privacy-preserving and federated machine learning in medical imaging.Nature Machine Intelligence, 2(6):305–311, 2020

    Georgios A Kaissis, Marcus R Makowski, Daniel R¨ uckert, and Rickmer F Braren. Secure, privacy-preserving and federated machine learning in medical imaging.Nature Machine Intelligence, 2(6):305–311, 2020

    2020

  4. [4]

    Advances and open problems in federated learning.Foundations and Trends® in Machine Learning, 14(1–2):1–210, 2021

    Peter Kairouz, H Brendan McMahan, Brendan Avent, Aur´ elien Bellet, Mehdi Bennis, Arjun Nitin Bhagoji, Kallista Bonawitz, Zachary Charles, Graham Cormode, Rachel Cum- mings, et al. Advances and open problems in federated learning.Foundations and Trends® in Machine Learning, 14(1–2):1–210, 2021

    2021

  5. [5]

    Federated learning: collaborative machine learning without centralized training data

    Abhishek V A, Binny S2, Johan T R, Nithin Raj, and Vishal Thomas. Federated learning: collaborative machine learning without centralized training data. volume 6, page 355, 2022

    2022

  6. [6]

    Measuring the Effects of Non-Identical Data Distribution for Federated Visual Classification

    Tzu-Ming Harry Hsu, Hang Qi, and Matthew Brown. Measuring the effects of non-identical data distribution for federated visual classification.arXiv preprint arXiv:1909.06335, 2019

    work page internal anchor Pith review Pith/arXiv arXiv 1909

  7. [7]

    Scaffold: Stochastic controlled averaging for federated learn- ing

    Sai Praneeth Karimireddy, Satyen Kale, Mehryar Mohri, Sashank Reddi, Sebastian Stich, and Ananda Theertha Suresh. Scaffold: Stochastic controlled averaging for federated learn- ing. InInternational conference on machine learning, pages 5132–5143. PMLR, 2020

    2020

  8. [8]

    Continual local training for better initialization of federated models

    Xin Yao and Lifeng Sun. Continual local training for better initialization of federated models. In2020 IEEE International Conference on Image Processing (ICIP), pages 1736–

  9. [9]

    Federated optimization in heterogeneous networks.Proceedings of Machine learning and systems, 2:429–450, 2020

    Tian Li, Anit Kumar Sahu, Manzil Zaheer, Maziar Sanjabi, Ameet Talwalkar, and Virginia Smith. Federated optimization in heterogeneous networks.Proceedings of Machine learning and systems, 2:429–450, 2020

    2020

  10. [10]

    Local learning matters: Rethinking data heterogeneity in federated learning

    Matias Mendieta, Taojiannan Yang, Pu Wang, Minwoo Lee, Zhengming Ding, and Chen Chen. Local learning matters: Rethinking data heterogeneity in federated learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8397–8406, 2022. 16 Accepted for Publication in Neurocomputing (2025)

    2022

  11. [11]

    Fedmd: Heterogenous federated learning via model distillation,

    Daliang Li and Junpu Wang. Fedmd: Heterogenous federated learning via model distilla- tion.arXiv preprint arXiv:1910.03581, 2019

    work page arXiv 1910

  12. [12]

    Ensemble attention distillation for privacy-preserving fed- erated learning

    Xuan Gong, Abhishek Sharma, Srikrishna Karanam, Ziyan Wu, Terrence Chen, David Doermann, and Arun Innanje. Ensemble attention distillation for privacy-preserving fed- erated learning. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 15076–15086, 2021

    2021

  13. [13]

    Fine-tuning global model via data-free knowledge distillation for non-iid federated learning

    Lin Zhang, Li Shen, Liang Ding, Dacheng Tao, and Ling-Yu Duan. Fine-tuning global model via data-free knowledge distillation for non-iid federated learning. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10174–10183, 2022

    2022

  14. [14]

    Reliable and inter- pretable personalized federated learning

    Zixuan Qin, Liu Yang, Qilong Wang, Yahong Han, and Qinghua Hu. Reliable and inter- pretable personalized federated learning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20422–20431, 2023

    2023

  15. [15]

    Pa- rameterized knowledge transfer for personalized federated learning.Advances in Neural Information Processing Systems, 34:10092–10104, 2021

    Jie Zhang, Song Guo, Xiaosong Ma, Haozhao Wang, Wenchao Xu, and Feijie Wu. Pa- rameterized knowledge transfer for personalized federated learning.Advances in Neural Information Processing Systems, 34:10092–10104, 2021

    2021

  16. [16]

    Layer-wised model aggregation for personalized federated learning

    Xiaosong Ma, Jie Zhang, Song Guo, and Wenchao Xu. Layer-wised model aggregation for personalized federated learning. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10092–10101, 2022

    2022

  17. [17]

    One-Shot Federated Learning

    Neel Guha, Ameet Talwalkar, and Virginia Smith. One-shot federated learning.arXiv preprint arXiv:1902.11175, 2019

    work page internal anchor Pith review Pith/arXiv arXiv 1902

  18. [18]

    Personalized federated learning with theoretical guarantees: A model-agnostic meta-learning approach.Advances in Neural Information Processing Systems, 33:3557–3568, 2020

    Alireza Fallah, Aryan Mokhtari, and Asuman Ozdaglar. Personalized federated learning with theoretical guarantees: A model-agnostic meta-learning approach.Advances in Neural Information Processing Systems, 33:3557–3568, 2020

    2020

  19. [19]

    Federated learning with data-agnostic distribution fusion

    Jian-hui Duan, Wenzhong Li, Derun Zou, Ruichen Li, and Sanglu Lu. Federated learning with data-agnostic distribution fusion. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8074–8083, 2023

    2023

  20. [20]

    Addressing class imbalance in federated learning

    Lixu Wang, Shichao Xu, Xiao Wang, and Qi Zhu. Addressing class imbalance in federated learning. InProceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 10165–10173, 2021

    2021

  21. [21]

    Federated visual classification with real-world data distribution

    Tzu-Ming Harry Hsu, Hang Qi, and Matthew Brown. Federated visual classification with real-world data distribution. InComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part X 16, pages 76–92. Springer, 2020

    2020

  22. [22]

    On bridging generic and personalized federated learn- ing for image classification.arXiv preprint arXiv:2107.00778, 2021

    Hong-You Chen and Wei-Lun Chao. On bridging generic and personalized federated learn- ing for image classification.arXiv preprint arXiv:2107.00778, 2021

    work page arXiv 2021

  23. [23]

    A review of methods for imbalanced multi-label classification.Pattern Recognition, 118:107965, 2021

    Adane Nega Tarekegn, Mario Giacobini, and Krzysztof Michalak. A review of methods for imbalanced multi-label classification.Pattern Recognition, 118:107965, 2021

    2021

  24. [24]

    Addressing fairness, bias and class imbalance in machine learning: the fbi-loss.arXiv preprint arXiv:2105.06345, 2021

    Elisa Ferrari and Davide Bacciu. Addressing fairness, bias and class imbalance in machine learning: the fbi-loss.arXiv preprint arXiv:2105.06345, 2021

    work page arXiv 2021

  25. [25]

    Handling inter-class and intra-class imbalance in class-imbalanced learning

    Zhining Liu, Pengfei Wei, Zhepei Wei, Boyang Yu, Jing Jiang, Wei Cao, Jiang Bian, and Yi Chang. Handling inter-class and intra-class imbalance in class-imbalanced learning. arXiv preprint arXiv:2111.12791, 2021. 17 Accepted for Publication in Neurocomputing (2025)

    work page arXiv 2021

  26. [26]

    Towards personalized federated learning.IEEE Transactions on Neural Networks and Learning Systems, 2022

    Alysa Ziying Tan, Han Yu, Lizhen Cui, and Qiang Yang. Towards personalized federated learning.IEEE Transactions on Neural Networks and Learning Systems, 2022

    2022

  27. [27]

    Ditto: Fair and robust feder- ated learning through personalization

    Tian Li, Shengyuan Hu, Ahmad Beirami, and Virginia Smith. Ditto: Fair and robust feder- ated learning through personalization. InInternational Conference on Machine Learning, pages 6357–6368. PMLR, 2021

    2021

  28. [28]

    Model-contrastive federated learning

    Qinbin Li, Bingsheng He, and Dawn Song. Model-contrastive federated learning. InPro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10713–10722, 2021

    2021

  29. [29]

    Gradaug: A new regularization method for deep neural networks.Advances in Neural Information Processing Systems, 33:14207– 14218, 2020

    Taojiannan Yang, Sijie Zhu, and Chen Chen. Gradaug: A new regularization method for deep neural networks.Advances in Neural Information Processing Systems, 33:14207– 14218, 2020

    2020

  30. [30]

    Isfl: Federated learning for non-iid data with local importance sampling.IEEE Internet of Things Journal, 2024

    Zheqi Zhu, Yuchen Shi, Pingyi Fan, Chenghui Peng, and Khaled B Letaief. Isfl: Federated learning for non-iid data with local importance sampling.IEEE Internet of Things Journal, 2024

    2024

  31. [31]

    Robust federated learning with noisy and heterogeneous clients

    Xiuwen Fang and Mang Ye. Robust federated learning with noisy and heterogeneous clients. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10072–10081, 2022

    2022

  32. [32]

    Fair resource allocation in federated learning.arXiv preprint arXiv:1905.10497, 2019

    Tian Li, Maziar Sanjabi, Ahmad Beirami, and Virginia Smith. Fair resource allocation in federated learning.arXiv preprint arXiv:1905.10497, 2019

    work page arXiv 1905

  33. [33]

    Tackling the objective inconsistency problem in heterogeneous federated optimization.Advances in neural information processing systems, 33:7611–7623, 2020

    Jianyu Wang, Qinghua Liu, Hao Liang, Gauri Joshi, and H Vincent Poor. Tackling the objective inconsistency problem in heterogeneous federated optimization.Advances in neural information processing systems, 33:7611–7623, 2020

    2020

  34. [34]

    Fedopt: Towards communication efficiency and privacy preservation in federated learning.Applied Sciences, 10(8):2864, 2020

    Muhammad Asad, Ahmed Moustafa, and Takayuki Ito. Fedopt: Towards communication efficiency and privacy preservation in federated learning.Applied Sciences, 10(8):2864, 2020

    2020

  35. [35]

    An agnostic approach to federated learning with class imbalance

    Zebang Shen, Juan Cervino, Hamed Hassani, and Alejandro Ribeiro. An agnostic approach to federated learning with class imbalance. InInternational Conference on Learning Rep- resentations, 2021

    2021

  36. [36]

    Federated learning with label distribution skew via logits calibration

    Jie Zhang, Zhiqi Li, Bo Li, Jianghe Xu, Shuang Wu, Shouhong Ding, and Chao Wu. Federated learning with label distribution skew via logits calibration. InInternational Conference on Machine Learning, pages 26311–26329. PMLR, 2022

    2022

  37. [37]

    Hyperfed: hyperbolic prototypes exploration with consistent aggregation for non-iid data in federated learning.arXiv preprint arXiv:2307.14384, 2023

    Xinting Liao, Weiming Liu, Chaochao Chen, Pengyang Zhou, Huabin Zhu, Yanchao Tan, Jun Wang, and Yue Qi. Hyperfed: hyperbolic prototypes exploration with consistent aggregation for non-iid data in federated learning.arXiv preprint arXiv:2307.14384, 2023

    work page arXiv 2023

  38. [38]

    Joint local relational augmentation and global nash equilibrium for federated learning with non-iid data

    Xinting Liao, Chaochao Chen, Weiming Liu, Pengyang Zhou, Huabin Zhu, Shuheng Shen, Weiqiang Wang, Mengling Hu, Yanchao Tan, and Xiaolin Zheng. Joint local relational augmentation and global nash equilibrium for federated learning with non-iid data. In Proceedings of the 31st ACM International Conference on Multimedia, pages 1536–1545, 2023

    2023

  39. [39]

    Federated probabilistic preference distribution modelling with compactness co-clustering for privacy-preserving multi-domain recommendation

    Weiming Liu, Chaochao Chen, Xinting Liao, Mengling Hu, Jianwei Yin, Yanchao Tan, and Longfei Zheng. Federated probabilistic preference distribution modelling with compactness co-clustering for privacy-preserving multi-domain recommendation. InProceedings of the 32rd International Joint Conference on Artificial Intelligence (IJCAI), pages 2206–2214, 2023. ...

    2023

  40. [40]

    Fediic: Towards robust federated learning for class-imbalanced medical image classification

    Nannan Wu, Li Yu, Xin Yang, Kwang-Ting Cheng, and Zengqiang Yan. Fediic: Towards robust federated learning for class-imbalanced medical image classification. InInternational Conference on Medical Image Computing and Computer-Assisted Intervention, pages 692–

  41. [41]

    Federated learning on heteroge- neous and long-tailed data via classifier re-training with federated features.arXiv preprint arXiv:2204.13399, 2022

    Xinyi Shang, Yang Lu, Gang Huang, and Hanzi Wang. Federated learning on heteroge- neous and long-tailed data via classifier re-training with federated features.arXiv preprint arXiv:2204.13399, 2022

    work page arXiv 2022

  42. [42]

    Robust representation learning with reliable pseudo-labels generation via self-adaptive optimal transport for short text clustering.arXiv preprint arXiv:2305.16335, 2023

    Xiaolin Zheng, Mengling Hu, Weiming Liu, Chaochao Chen, and Xinting Liao. Robust representation learning with reliable pseudo-labels generation via self-adaptive optimal transport for short text clustering.arXiv preprint arXiv:2305.16335, 2023

    work page arXiv 2023

  43. [43]

    Clip- guided federated learning on heterogeneity and long-tailed data

    Jiangming Shi, Shanshan Zheng, Xiangbo Yin, Yang Lu, Yuan Xie, and Yanyun Qu. Clip- guided federated learning on heterogeneity and long-tailed data. InProceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 14955–14963, 2024

    2024

  44. [44]

    Learning to model the tail.Advances in neural information processing systems, 30, 2017

    Yu-Xiong Wang, Deva Ramanan, and Martial Hebert. Learning to model the tail.Advances in neural information processing systems, 30, 2017

    2017

  45. [45]

    Sharing representations for long tail computer vision problems

    Samy Bengio. Sharing representations for long tail computer vision problems. InProceed- ings of the 2015 ACM on International Conference on Multimodal Interaction, pages 1–1, 2015

    2015

  46. [46]

    Factors in finetuning deep model for object detection with long-tail distribution

    Wanli Ouyang, Xiaogang Wang, Cong Zhang, and Xiaokang Yang. Factors in finetuning deep model for object detection with long-tail distribution. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 864–873, 2016

    2016

  47. [48]

    The Devil is in the Tails: Fine-grained Classification in the Wild

    Grant Van Horn and Pietro Perona. The devil is in the tails: Fine-grained classification in the wild.arXiv preprint arXiv:1709.01450, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  48. [49]

    Deep Active Learning over the Long Tail

    Yonatan Geifman and Ran El-Yaniv. Deep active learning over the long tail.arXiv preprint arXiv:1711.00941, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  49. [50]

    Range loss for deep face recognition with long-tailed training data

    Xiao Zhang, Zhiyuan Fang, Yandong Wen, Zhifeng Li, and Yu Qiao. Range loss for deep face recognition with long-tailed training data. InProceedings of the IEEE International Conference on Computer Vision, pages 5409–5418, 2017

    2017

  50. [51]

    Feature transfer learning for deep face recognition with under-represented data.arXiv preprint arXiv:1803.09014, 2018

    Xi Yin, Xiang Yu, Kihyuk Sohn, Xiaoming Liu, and Manmohan Chandraker. Feature transfer learning for deep face recognition with under-represented data.arXiv preprint arXiv:1803.09014, 2018

    work page arXiv 2018

  51. [52]

    A learning method for the class imbalance problem with medical data sets.Computers in biology and medicine, 40(5):509–518, 2010

    Der-Chiang Li, Chiao-Wen Liu, and Susan C Hu. A learning method for the class imbalance problem with medical data sets.Computers in biology and medicine, 40(5):509–518, 2010

    2010

  52. [53]

    A review on imbalanced data handling using undersampling and oversampling technique.Int

    Mayuri S Shelke, Prashant R Deshmukh, and Vijaya K Shandilya. A review on imbalanced data handling using undersampling and oversampling technique.Int. J. Recent Trends Eng. Res, 3(4):444–449, 2017

    2017

  53. [54]

    Imbalanced adversarial training with reweighting

    Wentao Wang, Han Xu, Xiaorui Liu, Yaxin Li, Bhavani Thuraisingham, and Jiliang Tang. Imbalanced adversarial training with reweighting. In2022 IEEE International Conference on Data Mining (ICDM), pages 1209–1214. IEEE, 2022. 19 Accepted for Publication in Neurocomputing (2025)

    2022

  54. [56]

    Addressing the class imbalance problem in medical image segmentation via accelerated tversky loss function

    Nikhil Nasalwai, Narinder Singh Punn, Sanjay Kumar Sonbhadra, and Sonali Agarwal. Addressing the class imbalance problem in medical image segmentation via accelerated tversky loss function. InPacific-Asia conference on knowledge discovery and data mining, pages 390–402. Springer, 2021

    2021

  55. [57]

    Distributional robustness loss for long-tail learning

    Dvir Samuel and Gal Chechik. Distributional robustness loss for long-tail learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 9495– 9504, 2021

    2021

  56. [58]

    Learning deep representation for imbalanced classification

    Chen Huang, Yining Li, Chen Change Loy, and Xiaoou Tang. Learning deep representation for imbalanced classification. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 5375–5384, 2016

    2016

  57. [59]

    Class-balanced loss based on effective number of samples

    Yin Cui, Menglin Jia, Tsung-Yi Lin, Yang Song, and Serge Belongie. Class-balanced loss based on effective number of samples. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9268–9277, 2019

    2019

  58. [60]

    https://www.kaggle.com/code/redwankarims- ony/nih-chest-x-ray8-classifier-cnn- visualization

    SONY. https://www.kaggle.com/code/redwankarims- ony/nih-chest-x-ray8-classifier-cnn- visualization. 2020

    2020

  59. [61]

    K.; Hayase, J.; Srinivasa, S

    Samuel K Ainsworth, Jonathan Hayase, and Srinivasa. Git re-basin: Merging models modulo permutation symmetries, 2022.URL https://arxiv. org/abs/2209.04836, 2022

    work page arXiv 2022

  60. [62]

    On the convergence of fedavg on non-iid data,

    Xiang Li, Kaixuan Huang, Wenhao Yang, Shusen Wang, and Zhihua Zhang. On the convergence of fedavg on non-iid data.arXiv preprint arXiv:1907.02189, 2019

    work page arXiv 1907

  61. [63]

    Chexpert: A large chest radiograph dataset with uncertainty labels and expert comparison

    Jeremy Irvin, Pranav Rajpurkar, Michael Ko, Yifan Yu, Silviana Ciurea-Ilcus, Chris Chute, Henrik Marklund, Behzad Haghgoo, Robyn Ball, Katie Shpanskaya, et al. Chexpert: A large chest radiograph dataset with uncertainty labels and expert comparison. InProceed- ings of the AAAI conference on artificial intelligence, volume 33, pages 590–597, 2019

    2019

  62. [64]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009

    2009

  63. [65]

    Imagenet large scale visual recognition challenge

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhi- heng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, et al. Imagenet large scale visual recognition challenge. InInternational journal of computer vision, volume 115, pages 211–252. Springer, 2015

    2015

  64. [66]

    A state-of-the-art survey on solving non-iid data in federated learning.Future Generation Computer Systems, 135:244–258, 2022

    Xiaodong Ma, Jia Zhu, Zhihao Lin, Shanxuan Chen, and Yangjie Qin. A state-of-the-art survey on solving non-iid data in federated learning.Future Generation Computer Systems, 135:244–258, 2022

    2022

  65. [67]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016

    2016

  66. [68]

    A survey on addressing high-class imbalance in big data

    JL Leevy, TM Khoshgoftaar, RA Bauder, and N Seliya. A survey on addressing high-class imbalance in big data. j big data 5 (1): 42. 2018

    2018

  67. [69]

    Bridging the gap be- tween natural and medical images through deep colorization

    Lia Morra, Luca Piano, Fabrizio Lamberti, and Tatiana Tommasi. Bridging the gap be- tween natural and medical images through deep colorization. In2020 25th International Conference on Pattern Recognition (ICPR), pages 835–842. IEEE, 2021. 20 Accepted for Publication in Neurocomputing (2025) Appendix A Foreword This section includes several detailed explan...

    2021

  68. [70]

    E||▽Fk(wk t , ξ)−▽F k(wk t )||2 ≤σ 2 k (16) This means that when we train the local model, the loss value has an upper bound

    assumes that the stochastic gradient has a limit and this value impacts the overall convergence result bound. E||▽Fk(wk t , ξ)−▽F k(wk t )||2 ≤σ 2 k (16) This means that when we train the local model, the loss value has an upper bound. In the case in which data are highly skewed, it is well known that the average loss value of the minority class is higher...

    2025