pith. sign in

arxiv: 2405.16240 · v3 · submitted 2024-05-25 · 💻 cs.LG

AFL: A Single-Round Analytic Approach for Federated Learning with Pre-trained Models

Run He , Kai Tong , Di Fang , Han Sun , Ziqian Zeng , Haoran Li , Tianyi Chen , Huiping Zhuang This is my paper

Pith reviewed 2026-05-24 01:10 UTC · model grok-4.3

classification 💻 cs.LG
keywords analytic federated learningpre-trained modelsdata partitioning invariancesingle-round aggregationclosed-form solutionsfederated learningabsolute aggregation law
0
0 comments X

The pith

Analytic federated learning produces identical models regardless of how data is split across clients.

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

The paper presents analytic federated learning as a method that replaces iterative training with closed-form solutions applied once per client to pre-trained models. These local solutions are then combined through an absolute aggregation law in a single round. The resulting model remains unchanged no matter how the complete dataset is divided among participating clients. This property removes the need for multiple communication rounds and addresses common problems with uneven data distributions and varying client counts. Experiments across non-IID conditions and large client groups show consistent competitive results.

Core claim

Analytic federated learning trains each client on a pre-trained model with a one-epoch closed-form solution and aggregates the outcomes via an absolute aggregation law in one round, yielding a final model that is invariant to the partitioning of the full dataset among clients.

What carries the argument

The absolute aggregation law, which directly combines closed-form local solutions from pre-trained models into a single-round result.

If this is right

  • Single-round aggregation eliminates multiple communication rounds and speeds convergence.
  • Performance remains stable under extremely non-IID data distributions.
  • Results hold with large numbers of clients such as one thousand or more.
  • The approach shows invariance to data heterogeneity and to the number of clients.

Where Pith is reading between the lines

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

  • The invariance property could simplify federated systems where client data availability changes over time.
  • Direct aggregation of closed-form solutions might apply to other distributed settings beyond standard federated learning.
  • One-epoch local training reduces local compute demands compared with multi-epoch methods.

Load-bearing premise

The closed-form analytic solution from each client's local training on the pre-trained model can be aggregated via the absolute aggregation law without any loss of correctness.

What would settle it

Apply the method to the same full dataset partitioned in two different ways among clients and check whether the final aggregated model parameters match exactly.

Figures

Figures reproduced from arXiv: 2405.16240 by Di Fang, Han Sun, Haoran Li, Huiping Zhuang, Kai Tong, Run He, Tianyi Chen, Ziqian Zeng.

Figure 1
Figure 1. Figure 1: An overview of the AFL. During the local stage, each client calculates [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Accuracy over various number of clients. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Accuracy curves with communication rounds. Average [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
read the original abstract

In this paper, we introduce analytic federated learning (AFL), a new training paradigm that brings analytical (i.e., closed-form) solutions to the federated learning (FL) with pre-trained models. Our AFL draws inspiration from analytic learning -- a gradient-free technique that trains neural networks with analytical solutions in one epoch. In the local client training stage, the AFL facilitates a one-epoch training, eliminating the necessity for multi-epoch updates. In the aggregation stage, we derive an absolute aggregation (AA) law. This AA law allows a single-round aggregation, reducing heavy communication overhead and achieving fast convergence by removing the need for multiple aggregation rounds. More importantly, the AFL exhibits a property that \textit{invariance to data partitioning}, meaning that regardless of how the full dataset is distributed among clients, the aggregated result remains identical. This could spawn various potentials, such as data heterogeneity invariance and client-number invariance. We conduct experiments across various FL settings including extremely non-IID ones, and scenarios with a large number of clients (e.g., $\ge 1000$). In all these settings, our AFL constantly performs competitively while existing FL techniques encounter various obstacles. Our codes are available at https://github.com/ZHUANGHP/Analytic-federated-learning.

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 / 3 minor

Summary. The paper introduces Analytic Federated Learning (AFL), a single-round federated learning method for pre-trained models. Local clients perform one-epoch analytic (closed-form) training; an absolute aggregation (AA) law then combines results in one communication round. The central claim is that the AA law yields results invariant to how the global dataset is partitioned across clients (including extreme non-IID and varying client counts), recovering the same solution as centralized analytic training.

Significance. If the invariance property holds exactly, AFL would eliminate iterative communication rounds and provide a partition-independent solution for linear heads on fixed features, offering clear practical gains in communication cost and robustness to heterogeneity. The experiments on large client counts and non-IID partitions support the practical utility if the math is exact.

major comments (2)
  1. [§3.2] §3.2, Eq. (8)–(10): the absolute aggregation law is stated to recover the global closed-form solution exactly by summing local Gram matrices and cross terms. A short explicit verification (or reference to the known property of sufficient statistics for linear least-squares) that the aggregated parameters equal the centralized pseudoinverse solution on the concatenated feature matrix would strengthen the central invariance claim.
  2. [§4.1] §4.1, the local analytic solution: the derivation assumes a fixed pre-trained feature extractor and solves only for the linear head. It is unclear whether the same AA law extends without approximation when the pre-trained backbone is also updated or when non-linear heads are used; this assumption is load-bearing for the “pre-trained models” scope.
minor comments (3)
  1. [Table 2] Table 2 and Figure 3: axis labels and legend entries use inconsistent abbreviations (e.g., “AFL” vs. “AnalyticFL”); standardize notation.
  2. [§5.3] §5.3: the claim of “client-number invariance” is supported by experiments up to 1000 clients, but the scaling plot would benefit from an explicit statement of wall-clock communication cost reduction relative to FedAvg.
  3. [Related Work] Related-work section: the connection to analytic learning (e.g., the cited gradient-free methods) is mentioned but lacks a direct comparison of per-client compute versus standard back-propagation.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the positive recommendation of minor revision and the constructive comments. We address each major comment below.

read point-by-point responses

  1. Referee: [§3.2] §3.2, Eq. (8)–(10): the absolute aggregation law is stated to recover the global closed-form solution exactly by summing local Gram matrices and cross terms. A short explicit verification (or reference to the known property of sufficient statistics for linear least-squares) that the aggregated parameters equal the centralized pseudoinverse solution on the concatenated feature matrix would strengthen the central invariance claim.

    Authors: We agree that an explicit verification would strengthen the central claim. In the revised manuscript we will add a short paragraph in §3.2 that directly shows the aggregated Gram matrix and cross-term vector are exactly the sufficient statistics for the global linear least-squares problem; therefore the closed-form solution obtained after aggregation is identical to the pseudoinverse solution computed on the concatenated feature matrix. revision: yes

  2. Referee: [§4.1] §4.1, the local analytic solution: the derivation assumes a fixed pre-trained feature extractor and solves only for the linear head. It is unclear whether the same AA law extends without approximation when the pre-trained backbone is also updated or when non-linear heads are used; this assumption is load-bearing for the “pre-trained models” scope.

    Authors: The AFL framework is explicitly scoped to pre-trained models with a frozen backbone and a linear head, as stated in the title, abstract, and §4.1. The exact AA law and partition invariance rely on the linearity of the head and the fixed features; the paper makes no claim that the same law holds without approximation for trainable backbones or non-linear heads. We will insert one clarifying sentence in §4.1 to restate this scope and the underlying assumptions. revision: partial

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained via sufficient statistics equivalence

full rationale

The AFL derives local closed-form solutions on fixed pre-trained features and an absolute aggregation law that sums local Gram matrices and cross terms. This exactly recovers the centralized pseudoinverse/least-squares solution on concatenated data, making partition invariance a direct algebraic consequence rather than a fitted or self-referential claim. No self-citation chain, ansatz smuggling, or renaming of known results is load-bearing; the result is externally verifiable as standard distributed linear regression and does not reduce to its inputs by definition.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities are stated. The central claim rests on the unshown derivation of the absolute aggregation law and the assumption that analytic learning extends directly to pre-trained models in the federated setting.

pith-pipeline@v0.9.0 · 5778 in / 1207 out tokens · 18616 ms · 2026-05-24T01:10:08.286062+00:00 · methodology

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.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Analytic Personalized Federated Meta-Learning

    cs.DC 2025-02 unverdicted novelty 7.0

    Proposes FedACnnL for analytic layer-wise DNN training in federated settings and pFedACnnL for analytic personalized meta-learning, claiming 83-99% training time reduction and 4-8% accuracy gains over baselines with S...

Reference graph

Works this paper leans on

56 extracted references · 56 canonical work pages · cited by 1 Pith paper

  1. [1]

    Fed- erated learning based on dynamic regularization

    Durmus Alp Emre Acar, Yue Zhao, Ramon Matas, Matthew Mattina, Paul Whatmough, and Venkatesh Saligrama. Fed- erated learning based on dynamic regularization. InInter- national Conference on Learning Representations, 2021. 6, 3

    work page 2021

  2. [2]

    Distributed distillation for on-device learning

    Ilai Bistritz, Ariana Mann, and Nicholas Bambos. Distributed distillation for on-device learning. InAdvances in Neural Information Processing Systems, pages 22593–22604. Curran Associates, Inc., 2020. 2

    work page 2020

  3. [3]

    On the importance and applicability of pre- training for federated learning

    Hong-You Chen, Cheng-Hao Tu, Ziwei Li, Han Wei Shen, and Wei-Lun Chao. On the importance and applicability of pre- training for federated learning. InThe Eleventh International Conference on Learning Representations, 2023. 6

    work page 2023

  4. [4]

    Randall E. Cline. Representations for the generalized inverse of a partitioned matrix.Journal of the Society for Industrial and Applied Mathematics, 12(3):588–600, 1964. 4, 1

    work page 1964

  5. [5]

    ImageNet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. ImageNet: A large-scale hierarchical image database. In2009 IEEE Conference on Computer Vision and Pattern Recognition, pages 248–255, 2009. 6

    work page 2009

  6. [6]

    An image is worth 16x16 words: Transformers for image recognition at scale, 2021

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at scale, 2021. 8

    work page 2021

  7. [7]

    Learning federated visual prompt in null space for mri reconstruction

    Chun-Mei Feng, Bangjun Li, Xinxing Xu, Yong Liu, Huazhu Fu, and Wangmeng Zuo. Learning federated visual prompt in null space for mri reconstruction. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 8064–8073, 2023. 1, 6, 4

    work page 2023

  8. [8]

    Pseudoin- verse learning algorithm for feedforward neural networks

    Ping Guo, Michael R Lyu, and NE Mastorakis. Pseudoin- verse learning algorithm for feedforward neural networks. Advances in Neural Networks and Applications, pages 321– 326, 2001. 2, 3

    work page 2001

  9. [9]

    Online knowledge dis- tillation via collaborative learning

    Qiushan Guo, Xinjiang Wang, Yichao Wu, Zhipeng Yu, Ding Liang, Xiaolin Hu, and Ping Luo. Online knowledge dis- tillation via collaborative learning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020. 2

    work page 2020

  10. [10]

    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 (CVPR), 2016. 6

    work page 2016

  11. [11]

    FedFisher: Leveraging Fisher information for one-shot fed- erated learning

    Divyansh Jhunjhunwala, Shiqiang Wang, and Gauri Joshi. FedFisher: Leveraging Fisher information for one-shot fed- erated learning. InProceedings of The 27th International Conference on Artificial Intelligence and Statistics, pages 1612–1620. PMLR, 2024. 4

    work page 2024

  12. [12]

    Secure, privacy-preserving and feder- ated 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 feder- ated machine learning in medical imaging.Nature Machine Intelligence, 2(6):305–311, 2020. 1

    work page 2020

  13. [13]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Vinod Nair, and Geoffrey Hinton. Learning multiple layers of features from tiny images. Technical report, University of Toronto, 2009. 6

    work page 2009

  14. [14]

    Blockchain-federated- learning and deep learning models for covid-19 detection us- ing ct imaging.IEEE Sensors Journal, 21(14):16301–16314,

    Rajesh Kumar, Abdullah Aman Khan, Jay Kumar, Za- kria, Noorbakhsh Amiri Golilarz, Simin Zhang, Yang Ting, Chengyu Zheng, and Wenyong Wang. Blockchain-federated- learning and deep learning models for covid-19 detection us- ing ct imaging.IEEE Sensors Journal, 21(14):16301–16314,

    work page

  15. [15]

    Preservation of the global knowledge by not- true distillation in federated learning

    Gihun Lee, Minchan Jeong, Yongjin Shin, Sangmin Bae, and Se-Young Yun. Preservation of the global knowledge by not- true distillation in federated learning. InAdvances in Neural Information Processing Systems, pages 38461–38474. Curran Associates, Inc., 2022. 6, 3

    work page 2022

  16. [16]

    Model-contrastive federated learning

    Qinbin Li, Bingsheng He, and Dawn Song. Model-contrastive federated learning. InProceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR), pages 10713–10722, 2021. 6, 3

    work page 2021

  17. [17]

    Fed- erated learning on non-iid data silos: An experimental study

    Qinbin Li, Yiqun Diao, Quan Chen, and Bingsheng He. Fed- erated learning on non-iid data silos: An experimental study. In2022 IEEE 38th International Conference on Data Engi- neering (ICDE), pages 965–978, 2022. 1

    work page 2022

  18. [18]

    Federated optimiza- tion in heterogeneous networks

    Tian Li, Anit Kumar Sahu, Manzil Zaheer, Maziar Sanjabi, Ameet Talwalkar, and Virginia Smith. Federated optimiza- tion in heterogeneous networks. InProceedings of Machine Learning and Systems, pages 429–450, 2020. 1, 2, 6, 3

    work page 2020

  19. [19]

    Feder- ated learning with position-aware neurons

    Xin-Chun Li, Yi-Chu Xu, Shaoming Song, Bingshuai Li, Yinchuan Li, Yunfeng Shao, and De-Chuan Zhan. Feder- ated learning with position-aware neurons. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10082–10091, 2022. 2

    work page 2022

  20. [20]

    Revisiting weighted aggregation in federated learning with neural net- works

    Zexi Li, Tao Lin, Xinyi Shang, and Chao Wu. Revisiting weighted aggregation in federated learning with neural net- works. InProceedings of the 40th International Conference on Machine Learning, pages 19767–19788. PMLR, 2023. 2, 6

    work page 2023

  21. [21]

    Ensemble distillation for robust model fusion in federated learning

    Tao Lin, Lingjing Kong, Sebastian U Stich, and Martin Jaggi. Ensemble distillation for robust model fusion in federated learning. InAdvances in Neural Information Processing Systems, pages 2351–2363. Curran Associates, Inc., 2020. 6

    work page 2020

  22. [22]

    Ensemble distillation for robust model fusion in federated learning

    Tao Lin, Lingjing Kong, Sebastian U Stich, and Martin Jaggi. Ensemble distillation for robust model fusion in federated learning. InAdvances in Neural Information Processing Systems, pages 2351–2363. Curran Associates, Inc., 2020. 2

    work page 2020

  23. [23]

    Locality sensitive sparse encoding for learning world models online

    Zichen Liu, Chao Du, Wee Sun Lee, and Min Lin. Locality sensitive sparse encoding for learning world models online. InThe Twelfth International Conference on Learning Repre- sentations, 2024. 4 9

    work page 2024

  24. [24]

    Re- ducing communication overhead in federated learning for pre-trained language models using parameter-efficient fine- tuning

    Shubham Malaviya, Manish Shukla, and Sachin Lodha. Re- ducing communication overhead in federated learning for pre-trained language models using parameter-efficient fine- tuning. InProceedings of The 2nd Conference on Lifelong Learning Agents, pages 456–469. PMLR, 2023. 1

    work page 2023

  25. [25]

    Communication- Efficient Learning of Deep Networks from Decentralized Data

    Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Aguera y Arcas. Communication- Efficient Learning of Deep Networks from Decentralized Data. InProceedings of the 20th International Conference on Artificial Intelligence and Statistics, pages 1273–1282. PMLR,

    work page

  26. [26]

    Tiny imagenet, 2017

    Mohammed Ali mnmoustafa. Tiny imagenet, 2017. 6

    work page 2017

  27. [27]

    Park and I

    J. Park and I. W. Sandberg. Universal approximation using radial-basis-function networks.Neural Computation, 3(2): 246–257, 1991. 3

    work page 1991

  28. [28]

    A federated learning based approach for loan defaults prediction

    Geet Shingi. A federated learning based approach for loan defaults prediction. In2020 International Conference on Data Mining Workshops (ICDMW), pages 362–368, 2020. 1

    work page 2020

  29. [29]

    Very deep con- volutional networks for large-scale image recognition, 2015

    Karen Simonyan and Andrew Zisserman. Very deep con- volutional networks for large-scale image recognition, 2015. 8

    work page 2015

  30. [30]

    Fedproto: Federated prototype learning across heterogeneous clients.Proceedings of the AAAI Conference on Artificial Intelligence, 36(8):8432– 8440, 2022

    Yue Tan, Guodong Long, LU LIU, Tianyi Zhou, Qinghua Lu, Jing Jiang, and Chengqi Zhang. Fedproto: Federated prototype learning across heterogeneous clients.Proceedings of the AAAI Conference on Artificial Intelligence, 36(8):8432– 8440, 2022. 1

    work page 2022

  31. [31]

    Federated learning from pre-trained models: A contrastive learning approach

    Yue Tan, Guodong Long, Jie Ma, LU LIU, Tianyi Zhou, and Jing Jiang. Federated learning from pre-trained models: A contrastive learning approach. InAdvances in Neural In- formation Processing Systems, pages 19332–19344. Curran Associates, Inc., 2022. 1

    work page 2022

  32. [32]

    Learning from the kernel and the range space

    Kar-Ann Toh. Learning from the kernel and the range space. In2018 IEEE/ACIS 17th International Conference on Com- puter and Information Science (ICIS), pages 1–6, 2018. 3

    work page 2018

  33. [33]

    Learning from the kernel and the range space

    Kar-Ann Toh. Learning from the kernel and the range space. Inthe Proceedings of the 17th 2018 IEEE Conference on Computer and Information Science, pages 417–422. IEEE,

    work page 2018

  34. [34]

    DaFKD: Domain-aware federated knowledge distillation

    Haozhao Wang, Yichen Li, Wenchao Xu, Ruixuan Li, Yufeng Zhan, and Zhigang Zeng. DaFKD: Domain-aware federated knowledge distillation. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR), pages 20412–20421, 2023. 2

    work page 2023

  35. [35]

    FedCDA: Federated learning with cross-rounds divergence-aware aggregation

    Haozhao Wang, Haoran Xu, Yichen Li, Yuan Xu, Ruixuan Li, and Tianwei Zhang. FedCDA: Federated learning with cross-rounds divergence-aware aggregation. InThe Twelfth International Conference on Learning Representations, 2024. 2

    work page 2024

  36. [36]

    Vincent Poor

    Jianyu Wang, Qinghua Liu, Hao Liang, Gauri Joshi, and H. Vincent Poor. Tackling the objective inconsistency prob- lem in heterogeneous federated optimization. InAdvances in Neural Information Processing Systems, pages 7611–7623. Curran Associates, Inc., 2020. 1, 2

    work page 2020

  37. [37]

    DensePILAE: a feature reuse pseudoinverse learning algorithm for deep stacked au- toencoder.Complex & Intelligent Systems, pages 1–11, 2021

    Jue Wang, Ping Guo, and Yanjun Li. DensePILAE: a feature reuse pseudoinverse learning algorithm for deep stacked au- toencoder.Complex & Intelligent Systems, pages 1–11, 2021. 3

    work page 2021

  38. [38]

    X. Wang, T. Zhang, and R. Wang. Noniterative deep learning: Incorporating restricted boltzmann machine into multilayer random weight neural networks.IEEE Transactions on Sys- tems, Man, and Cybernetics: Systems, 49(7):1299–1308, 2019. 3

    work page 2019

  39. [39]

    Peer collaborative learning for online knowledge distillation.Proceedings of the AAAI Conference on Artificial Intelligence, 35(12):10302–10310,

    Guile Wu and Shaogang Gong. Peer collaborative learning for online knowledge distillation.Proceedings of the AAAI Conference on Artificial Intelligence, 35(12):10302–10310,

    work page

  40. [40]

    Advancing cross-domain dis- criminability in continual learning of vision-language models

    Yicheng Xu, Yuxin Chen, Jiahao Nie, Yusong Wang, Huiping Zhuang, and Manabu Okumura. Advancing cross-domain dis- criminability in continual learning of vision-language models. InThe Thirty-eighth Annual Conference on Neural Informa- tion Processing Systems, 2024. 8

    work page 2024

  41. [41]

    Ef- ficient model personalization in federated learning via client- specific prompt generation

    Fu-En Yang, Chien-Yi Wang, and Yu-Chiang Frank Wang. Ef- ficient model personalization in federated learning via client- specific prompt generation. InProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 19159–19168, 2023. 1

    work page 2023

  42. [42]

    Ffd: A federated learning based method for credit card fraud detection

    Wensi Yang, Yuhang Zhang, Kejiang Ye, Li Li, and Cheng- Zhong Xu. Ffd: A federated learning based method for credit card fraud detection. InBig Data – BigData 2019, pages 18–32, Cham, 2019. Springer International Publishing. 1

    work page 2019

  43. [43]

    FedDisco: Federated learning with discrepancy-aware collaboration

    Rui Ye, Mingkai Xu, Jianyu Wang, Chenxin Xu, Siheng Chen, and Yanfeng Wang. FedDisco: Federated learning with discrepancy-aware collaboration. InProceedings of the 40th International Conference on Machine Learning, pages 39879– 39902. PMLR, 2023. 6, 3

    work page 2023

  44. [44]

    Fed2: Feature- aligned federated learning

    Fuxun Yu, Weishan Zhang, Zhuwei Qin, Zirui Xu, Di Wang, Chenchen Liu, Zhi Tian, and Xiang Chen. Fed2: Feature- aligned federated learning. InProceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining, pages 2066–2074, 2021. 2

    work page 2066

  45. [45]

    PFLlib: Personal- ized federated learning algorithm library.arXiv preprint arXiv:2312.04992, 2023

    Jianqing Zhang, Yang Liu, Yang Hua, Hao Wang, Tao Song, Zhengui Xue, Ruhui Ma, and Jian Cao. PFLlib: Personal- ized federated learning algorithm library.arXiv preprint arXiv:2312.04992, 2023. 3

    work page arXiv 2023

  46. [46]

    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 (CVPR), pages 10174–10183, 2022. 2

    work page 2022

  47. [47]

    Fedpetuning: When feder- ated learning meets the parameter-efficient tuning methods of pre-trained language models

    Zhuo Zhang, Yuanhang Yang, Yong Dai, Qifan Wang, Yue Yu, Lizhen Qu, and Zenglin Xu. Fedpetuning: When feder- ated learning meets the parameter-efficient tuning methods of pre-trained language models. InFindings of the Association for Computational Linguistics: ACL 2023, page 9963–9977. Association for Computational Linguistics (ACL), 2023. An- nual Meetin...

    work page 2023

  48. [48]

    Data-free knowledge distillation for heterogeneous federated learning

    Zhuangdi Zhu, Junyuan Hong, and Jiayu Zhou. Data-free knowledge distillation for heterogeneous federated learning. InProceedings of the 38th International Conference on Ma- chine Learning, pages 12878–12889. PMLR, 2021. 6

    work page 2021

  49. [49]

    Data-free knowledge distillation for heterogeneous federated learning

    Zhuangdi Zhu, Junyuan Hong, and Jiayu Zhou. Data-free knowledge distillation for heterogeneous federated learning. 10 InProceedings of the 38th International Conference on Ma- chine Learning, pages 12878–12889. PMLR, 2021. 2

    work page 2021

  50. [50]

    Blockwise recursive Moore-Penrose inverse for network learning.IEEE Transactions on Systems, Man, and Cybernetics: Systems, pages 1–14, 2021

    Huiping Zhuang, Zhiping Lin, and Kar-Ann Toh. Blockwise recursive Moore-Penrose inverse for network learning.IEEE Transactions on Systems, Man, and Cybernetics: Systems, pages 1–14, 2021. 2, 3, 6

    work page 2021

  51. [51]

    Correlation projection for analytic learning of a classification network

    Huiping Zhuang, Zhiping Lin, and Kar-Ann Toh. Correlation projection for analytic learning of a classification network. Neural Processing Letters, pages 1–22, 2021. 8

    work page 2021

  52. [52]

    ACIL: Analytic class- incremental learning with absolute memorization and privacy protection

    Huiping Zhuang, Zhenyu Weng, Hongxin Wei, Renchunzi Xie, Kar-Ann Toh, and Zhiping Lin. ACIL: Analytic class- incremental learning with absolute memorization and privacy protection. InAdvances in Neural Information Processing Systems, pages 11602–11614. Curran Associates, Inc., 2022. 3, 4

    work page 2022

  53. [53]

    GKEAL: Gaussian kernel embedded analytic learning for few-shot class incremental task

    Huiping Zhuang, Zhenyu Weng, Run He, Zhiping Lin, and Ziqian Zeng. GKEAL: Gaussian kernel embedded analytic learning for few-shot class incremental task. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 7746–7755, 2023. 3

    work page 2023

  54. [54]

    GACL: Exemplar- free generalized analytic continual learning

    Huiping Zhuang, Yizhu Chen, Di Fang, Run He, Kai Tong, Hongxin Wei, Ziqian Zeng, and Cen Chen. GACL: Exemplar- free generalized analytic continual learning. InAdvances in Neural Information Processing Systems. Curran Associates, Inc., 2024. 3, 6

    work page 2024

  55. [55]

    DS-AL: A dual-stream analytic learning for exemplar-free class-incremental learning.Proceedings of the AAAI Conference on Artificial Intelligence, 38(15):17237– 17244, 2024

    Huiping Zhuang, Run He, Kai Tong, Ziqian Zeng, Cen Chen, and Zhiping Lin. DS-AL: A dual-stream analytic learning for exemplar-free class-incremental learning.Proceedings of the AAAI Conference on Artificial Intelligence, 38(15):17237– 17244, 2024. 3

    work page 2024

  56. [56]

    F-OAL: Forward- only online analytic learning with fast training and low mem- ory footprint in class incremental learning

    Huiping Zhuang, Yuchen Liu, Run He, Kai Tong, Ziqian Zeng, Cen Chen, Yi Wang, and Lap-Pui Chau. F-OAL: Forward- only online analytic learning with fast training and low mem- ory footprint in class incremental learning. InThe Thirty- eighth Annual Conference on Neural Information Processing Systems, 2024. 6 11 AFL: A Single-Round Analytic Approach for Fede...

    work page 2024