Federated Learning with Nonvacuous Generalisation Bounds

Benjamin Guedj; Maxime Haddouche; Pierre Jobic

arxiv: 2310.11203 · v2 · pith:PQ2NDQA7new · submitted 2023-10-17 · 💻 cs.LG · stat.ML

Federated Learning with Nonvacuous Generalisation Bounds

Pierre Jobic , Maxime Haddouche , Benjamin Guedj This is my paper

Pith reviewed 2026-05-24 05:47 UTC · model grok-4.3

classification 💻 cs.LG stat.ML

keywords federated learningPAC-Bayesian boundsgeneralization boundsprivacy preservationrandomized predictorsnonvacuous boundsdistributed learningsynchronous and heterogeneous settings

0 comments

The pith

Federated nodes train local private predictors whose combination yields a global predictor with nonvacuous PAC-Bayesian bounds.

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

The paper shows how multiple nodes can each train a randomized predictor on their own private data and release only that predictor. These local predictors are then assembled into one global randomized predictor that carries forward the same style of generalization bound. Experiments compare the approach against the unrealistic case where every node pools its full dataset, finding nearly matching accuracy together with explicit numerical bounds that remain non-vacuous. The work also quantifies the exact performance and bound penalty incurred by keeping data local rather than shared. This matters because it supplies a concrete route to privacy-preserving distributed learning that still comes with usable guarantees on future error.

Core claim

A global randomised predictor can be built from local private predictors in federated learning such that it inherits the PAC-Bayesian generalisation properties of the locals whenever each node optimises an objective derived from the bound; the construction works in the synchronous case, the heterogeneous case, and the homogeneous case, produces predictive performance comparable to the batch setting in which all datasets are shared, supplies numerically nonvacuous bounds, and allows explicit calculation of the performance and bound increments required to preserve privacy.

What carries the argument

The global randomised predictor that inherits PAC-Bayesian generalisation bounds from local private predictors, each trained by optimising an objective derived from the bound.

If this is right

In the synchronous case every node uses the identical bound-derived objective, so the global predictor directly inherits the local bounds.
In heterogeneous and homogeneous cases each node may use its own personalised objective yet the global predictor still inherits valid bounds.
Predictive performance stays comparable to the full-data batch baseline, with the exact increment in error and in bound value computed for the privacy-preserving versions.
Privacy is preserved because each node releases only its local predictor and never its training data.

Where Pith is reading between the lines

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

The same inheritance mechanism could be tested when local predictors are obtained by other privacy techniques that still produce randomized outputs.
If the bound increment grows slower than linearly with the number of nodes, the method would become relatively cheaper on very large networks.
The explicit price-of-privacy numbers could be used to decide, for a given accuracy target, whether federated training is preferable to centralised training under privacy constraints.
The approach suggests checking whether the nonvacuous bounds remain informative when the local datasets are much smaller or more unbalanced than those used in the reported experiments.

Load-bearing premise

The global randomised predictor inherits the PAC-Bayesian generalisation properties of the local private predictors when each node optimises an objective derived from the bound.

What would settle it

A run in which the measured error of the global predictor on unseen data substantially exceeds the inherited bound while each local predictor still satisfies its own bound.

Figures

Figures reproduced from arXiv: 2310.11203 by Benjamin Guedj, Maxime Haddouche, Pierre Jobic.

read the original abstract

We introduce a novel strategy to train randomised predictors in federated learning, where each node of the network aims at preserving its privacy by releasing a local predictor but keeping secret its training dataset with respect to the other nodes. We then build a global randomised predictor which inherits the properties of the local private predictors in the sense of a PAC-Bayesian generalisation bound. We consider the synchronous case where all nodes share the same training objective (derived from a generalisation bound), and the heterogenous and homogenous cases where each node may have its own personalised training objective. We show through a series of numerical experiments that our approach achieves a comparable predictive performance to that of the batch approach where all datasets are shared across nodes. Moreover the predictors are supported by numerically nonvacuous generalisation bounds while preserving privacy for each node. We explicitly compute the increment on predictive performance and generalisation bounds for our two federated settings, highlighting the price to pay to preserve privacy.

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.

Desk Editor's Note private letter to a colleague

Paper gets nonvacuous PAC-Bayes bounds in federated learning via local bound-derived objectives, but the global inheritance step needs a close look for extra looseness.

read the letter

The one or two things to know: this paper trains randomized predictors in federated learning so that each node keeps its data private, optimizes a local objective from a PAC-Bayes bound, and then combines them into a global predictor with a nonvacuous bound. They show in experiments that predictive performance is comparable to the centralized batch case, and the bounds stay numerically nonvacuous in synchronous, heterogeneous, and homogeneous settings. They also calculate the increment in performance loss and bound looseness due to privacy. What is new is the specific strategy of deriving the training objective from the generalization bound at each node and then claiming the global inherits it. The experiments back up the performance claim with concrete numbers. The soft spots are around the bound inheritance. The central assumption is that the global randomized predictor gets the PAC-Bayesian properties from the local ones. In the heterogeneous case where objectives differ, or even in synchronous, the aggregation step may require an extra inequality that adds looseness not accounted for in the local bounds. If the global KL term isn't bounded by the sum of locals without extra factors, the nonvacuous claim could weaken. The paper seems to rely on the experiments to show it works numerically, but the theoretical transfer needs to be tight for the claim to hold without qualification. This is aimed at the federated learning community, especially those wanting privacy-preserving methods with guarantees. Someone working on applying PAC-Bayes to distributed settings would get value from the construction and the numbers. It deserves a serious referee because the idea is concrete, the experiments are there, and the potential flaw in bound transfer is checkable with the full derivations.

Referee Report

2 major / 1 minor

Summary. The manuscript presents a federated learning approach for training randomized predictors where each node optimizes a local objective derived from a PAC-Bayesian generalization bound on its private data. A global randomized predictor is then formed that inherits the PAC-Bayesian properties. The work addresses synchronous, heterogeneous, and homogeneous settings, with experiments claiming comparable predictive performance to centralized batch learning, numerically nonvacuous bounds, and privacy preservation, while quantifying the performance and bound increments due to the federated setup.

Significance. Should the inheritance of the bounds hold rigorously without substantial additional looseness, the result would be significant for the field of privacy-preserving machine learning. Numerically nonvacuous PAC-Bayesian bounds in a federated context are a strong point, as most such bounds are vacuous. The explicit calculation of the 'price to pay' for privacy is a useful contribution if supported by the derivations.

major comments (2)

[Global predictor construction (likely §3)] The central construction asserts that the global randomized predictor inherits the PAC-Bayesian generalization properties of the local private predictors. However, the derivation must explicitly show how the global risk and KL-divergence terms are controlled by the local ones (e.g., whether the global KL is bounded by the sum of local KLs in the synchronous case, or how differing objectives affect the terms in the heterogeneous case). Without this, the transfer of numerical nonvacuousness from local to global bounds is not guaranteed.
[Experiments section] Table or figure reporting the generalization bounds (likely in the experiments section): the manuscript should include the precise numerical values of the PAC-Bayesian bounds for the global predictor, local predictors, and the batch baseline, along with any multiplicative or additive looseness factors introduced by the aggregation step, to allow verification that the bounds remain nonvacuous after inheritance.

minor comments (1)

[Notation and definitions] Clarify the exact form of the global prior and how it relates to the local priors used in each node's objective.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. We address the two major comments below and will revise the manuscript to strengthen the presentation of the bound inheritance and the experimental reporting.

read point-by-point responses

Referee: [Global predictor construction (likely §3)] The central construction asserts that the global randomized predictor inherits the PAC-Bayesian generalization properties of the local private predictors. However, the derivation must explicitly show how the global risk and KL-divergence terms are controlled by the local ones (e.g., whether the global KL is bounded by the sum of local KLs in the synchronous case, or how differing objectives affect the terms in the heterogeneous case). Without this, the transfer of numerical nonvacuousness from local to global bounds is not guaranteed.

Authors: We agree that the control of the global risk and KL terms should be stated more explicitly. In the revised manuscript we will expand the relevant section (currently §3) with a dedicated lemma that derives the global risk bound directly from the local risks and shows that the global KL divergence is at most the sum of the local KL divergences in the synchronous case; for the heterogeneous case we will add the corresponding weighted combination that arises from the personalised objectives. This will make the transfer of nonvacuousness fully rigorous and transparent. revision: yes
Referee: [Experiments section] Table or figure reporting the generalization bounds (likely in the experiments section): the manuscript should include the precise numerical values of the PAC-Bayesian bounds for the global predictor, local predictors, and the batch baseline, along with any multiplicative or additive looseness factors introduced by the aggregation step, to allow verification that the bounds remain nonvacuous after inheritance.

Authors: We will add a new table (or augmented existing table) in the experiments section that lists the exact numerical PAC-Bayesian bound values for the global predictor, each local predictor, and the centralized batch baseline. The table will also report the multiplicative and additive looseness factors that appear in the aggregation step, together with the observed gap between local and global bounds. This will allow direct verification that the inherited bounds remain nonvacuous. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper constructs local posteriors by optimizing objectives derived from PAC-Bayesian bounds on private node data, then asserts that a global randomized predictor inherits the generalization properties. The abstract and description present this inheritance as following from the synchronous/homogeneous/heterogeneous constructions without exhibiting a reduction by construction (e.g., no equation showing the global bound equals the sum of local bounds tautologically). No self-citation load-bearing steps, uniqueness theorems, or ansatz smuggling are identifiable from the provided text. Numerical experiments supply independent empirical checks on performance and bound values. The derivation chain remains self-contained relative to standard PAC-Bayesian theory.

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.

pith-pipeline@v0.9.0 · 5692 in / 1043 out tokens · 20483 ms · 2026-05-24T05:47:24.614800+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

73 extracted references · 73 canonical work pages · 5 internal anchors

[1]

Achituve, I., Shamsian, A., Navon, A., Chechik, G., and Fetaya, E. (2021). Personalized Federated Learning With Gaussian Processes . In Ranzato, M., Beygelzimer, A., Dauphin, Y. N., Liang, P., and Vaughan, J. W., editors, Advances in Neural Information Processing Systems 34: Annual Conference on Neural Information Processing Systems 2021, NeurIPS 2021, De...

work page 2021
[2]

T., Yu, F

Agarwal, N., Suresh, A. T., Yu, F. X., Kumar, S., and McMahan, B. (2018). cpSGD: Communication-efficient and differentially-private distributed SGD . In Bengio, S., Wallach, H. M., Larochelle, H., Grauman, K., Cesa - Bianchi, N., and Garnett, R., editors, Advances in Neural Information Processing Systems 31: Annual Conference on Neural Information Process...

work page 2018
[3]

Alquier, P. (2021). User-friendly introduction to PAC-Bayes bounds . arXiv , abs/2110.11216

work page arXiv 2021
[4]

Alquier, P., Ridgway, J., and Chopin, N. (2016). On the properties of variational approximations of G ibbs posteriors . Journal of Machine Learning Research

work page 2016
[5]

and Meir, R

Amit, R. and Meir, R. (2018). Meta-Learning by Adjusting Priors Based on Extended PAC-Bayes Theory . In International Conference on Machine Learning ( ICML )

work page 2018
[6]

Flower: A Friendly Federated Learning Research Framework

Beutel, D. J., Topal, T., Mathur, A., Qiu, X., Fernandez-Marques, J., Gao, Y., Sani, L., Kwing, H. L., Parcollet, T., Gusmão, P. P. d., and Lane, N. D. (2020). Flower: A Friendly Federated Learning Research Framework . arXiv preprint arXiv:2007.14390

work page internal anchor Pith review Pith/arXiv arXiv 2020
[7]

and Guedj, B

Biggs, F. and Guedj, B. (2021). Differentiable PAC-Bayes objectives with partially aggregated neural networks. Entropy , 23(10)

work page 2021
[8]

and Guedj, B

Biggs, F. and Guedj, B. (2022). Non-vacuous generalisation bounds for shallow neural networks. In Proceedings of the 39th International Conference on Machine Learning [ICML] , volume 162 of Proceedings of Machine Learning Research , pages 1963--1981. PMLR

work page 2022
[9]

Biggs, F., Zantedeschi, V., and Guedj, B. (2022). On margins and generalisation for voting classifiers. In NeurIPS

work page 2022
[10]

B., Patel, S., Ramage, D., Segal, A., and Seth, K

Bonawitz, K., Ivanov, V., Kreuter, B., Marcedone, A., McMahan, H. B., Patel, S., Ramage, D., Segal, A., and Seth, K. (2017). Practical secure aggregation for privacy-preserving machine learning. In Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security , pages 1175--1191

work page 2017
[11]

Boucheron, S., Lugosi, G., and Massart, P. (2013). Concentration Inequalities - A Nonasymptotic Theory of Independence . Oxford University Press

work page 2013
[12]

Catoni, O. (2007). PAC- Bayesian supervised classification: the thermodynamics of statistical learning . Institute of Mathematical Statistics

work page 2007
[13]

and Chao, W

Chen, H. and Chao, W. (2021). FedBE: Making Bayesian Model Ensemble Applicable to Federated Learning . In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021 . OpenReview.net

work page 2021
[14]

Ch\'erief-Abdellatif, B.-E., Shi, Y., Doucet, A., and Guedj, B. (2022). On PAC-Bayesian reconstruction guarantees for VAEs . In Proceedings of The 25th International Conference on Artificial Intelligence and Statistics [AISTATS] , volume 151 of Proceedings of Machine Learning Research , pages 3066--3079. PMLR

work page 2022
[15]

Chugg, B., Wang, H., and Ramdas, A. (2023). A unified recipe for deriving (time-uniform) PAC-B ayes bounds . arXiv , abs/2302.03421

work page arXiv 2023
[16]

Ding, N., Chen, X., Levinboim, T., Goodman, S., and Soricut, R. (2021). Bridging the Gap Between Practice and PAC-B ayes Theory in Few-Shot Meta-Learning . In Conference on Neural Information Processing Systems (NeurIPS)

work page 2021
[17]

Duchi, J. (2007). Derivations for linear algebra and optimization . Berkeley, California , 3(1):2325--5870

work page 2007
[18]

Dziugaite, G. K. and Roy, D. (2017). Computing Nonvacuous Generalization Bounds for Deep (Stochastic) Neural Networks with Many More Parameters than Training Data . In Conference on Uncertainty in Artificial Intelligence ( UAI )

work page 2017
[19]

Fard, M. M. and Pineau, J. (2010). PAC-Bayesian model selection for reinforcement learning. In Conference on Neural Information Processing Systems (NeurIPS)

work page 2010
[20]

and Majumdar, A

Farid, A. and Majumdar, A. (2021). Generalization Bounds for Meta-Learning via PAC-B ayes and Uniform Stability . In Conference on Neural Information Processing Systems (NeurIPS)

work page 2021
[21]

Guedj, B. (2019). A Primer on PAC-Bayesian Learning . In Proceedings of the second congress of the French Mathematical Society , volume 33

work page 2019
[22]

and Guedj, B

Haddouche, M. and Guedj, B. (2022). Online PAC-Bayes Learning . In Conference on Neural Information Processing Systems (NeurIPS)

work page 2022
[23]

and Guedj, B

Haddouche, M. and Guedj, B. (2023). PAC-Bayes Generalisation Bounds for Heavy-Tailed Losses through Supermartingales . Transactions on Machine Learning Research

work page 2023
[24]

Hardt, M., Price, E., and Srebro, N. (2016). Equality of Opportunity in Supervised Learning . In Lee, D. D., Sugiyama, M., von Luxburg, U., Guyon, I., and Garnett, R., editors, Advances in Neural Information Processing Systems 29: Annual Conference on Neural Information Processing Systems 2016, December 5-10, 2016, Barcelona, Spain , pages 3315--3323

work page 2016
[25]

Hellström, F., Durisi, G., Guedj, B., and Raginsky, M. (2023). Generalization Bounds: Perspectives from Information Theory and PAC-Bayes . arXiv

work page 2023
[26]

Jang, K., Jun, K.-S., Kuzborskij, I., and Orabona, F. (2023). Tighter PAC-Bayes Bounds Through Coin-Betting . arXiv , abs/2302.05829

work page arXiv 2023
[27]

B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A

Kairouz, P., McMahan, H. B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A. N., Bonawitz, K., Charles, Z., Cormode, G., Cummings, R., D'Oliveira, R. G. L., Eichner, H., Rouayheb, S. E., Evans, D., Gardner, J., Garrett, Z., Gasc \'o n, A., Ghazi, B., Gibbons, P. B., Gruteser, M., Harchaoui, Z., He, C., He, L., Huo, Z., Hutchinson, B., Hsu, J., Jaggi, M., J...

work page 2021
[28]

and Hospedales, T

Kim, M. and Hospedales, T. M. (2023). FedHB: Hierarchical Bayesian Federated Learning . arXiv , abs/2305.04979

work page arXiv 2023
[29]

Kingma, D. P. and Welling, M. (2014). Auto-encoding variational bayes. In Bengio, Y. and LeCun, Y., editors, 2nd International Conference on Learning Representations, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings

work page 2014
[30]

Federated Optimization: Distributed Machine Learning for On-Device Intelligence

Kone c n \'y , J., McMahan, H. B., Ramage, D., and Richt \' a rik, P. (2016a). Federated Optimization: Distributed Machine Learning for On-Device Intelligence . arXiv , abs/1610.02527

work page internal anchor Pith review Pith/arXiv arXiv
[31]

Federated Learning: Strategies for Improving Communication Efficiency

Kone c n \'y , J., McMahan, H. B., Yu, F. X., Richt \' a rik, P., Suresh, A. T., and Bacon, D. (2016b). Federated Learning: Strategies for Improving Communication Efficiency . arXiv , abs/1610.05492

work page internal anchor Pith review Pith/arXiv arXiv
[32]

Kotelevskii, N., Vono, M., Durmus, A., and Moulines, E. (2022). FedPop: A Bayesian Approach for Personalised Federated Learning . In NeurIPS

work page 2022
[33]

and Szepesvári, C

Kuzborskij, I. and Szepesvári, C. (2019). Efron- S tein PAC - B ayesian I nequalities. arXiv:1909.01931

work page arXiv 2019
[34]

Letarte, G., Germain, P., Guedj, B., and Laviolette, F. (2019). Dichotomize and generalize: PAC-Bayesian binary activated deep neural networks. In Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems [NeurIPS] 2019, 8-14 December 2019, Vancouver, BC, Canada , pages 6869--6879

work page 2019
[35]

Li, L., Guedj, B., and Loustau, S. (2018). A quasi- Bayesian perspective to online clustering. Electron. J. Statist. , 12(2):3071--3113

work page 2018
[36]

Mammen, P. M. (2021). Federated Learning: Opportunities and Challenges

work page 2021
[37]

Maurer, A. (2004). A note on the PAC-Bayesian theorem . arXiv , cs/0411099

work page internal anchor Pith review Pith/arXiv arXiv 2004
[38]

McAllester, D. A. (1998). Some PAC-Bayesian theorems . In Proceedings of the eleventh annual conference on Computational Learning Theory , pages 230--234. ACM

work page 1998
[39]

McAllester, D. A. (2003). PAC-Bayesian stochastic model selection . Machine Learning , 51(1):5--21

work page 2003
[40]

McMahan, B., Moore, E., Ramage, D., Hampson, S., and Arcas, B. A. y. (2017a). Communication-Efficient Learning of Deep Networks from Decentralized Data . In Singh, A. and Zhu, J., editors, Proceedings of the 20th International Conference on Artificial Intelligence and Statistics , volume 54 of Proceedings of Machine Learning Research , pages 1273--1282. PMLR

work page
[41]

B., Moore, E., Ramage, D., Hampson, S., and y Arcas, B

McMahan, H. B., Moore, E., Ramage, D., Hampson, S., and y Arcas, B. A. (2017b). Communication-Efficient Learning of Deep Networks from Decentralized Data

work page
[42]

Mhammedi, Z., Gr \" u nwald, P., and Guedj, B. (2019). PAC-Bayes un-expected Bernstein inequality. In Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems [NeurIPS] 2019, 8-14 December 2019, Vancouver, BC, Canada , pages 12180--12191

work page 2019
[43]

Mohri, M., Sivek, G., and Suresh, A. T. (2019). Agnostic Federated Learning . In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning , volume 97 of Proceedings of Machine Learning Research , pages 4615--4625. PMLR

work page 2019
[44]

Nozawa, K., Germain, P., and Guedj, B. (2020). PAC-Bayesian contrastive unsupervised representation learning. In Conference on Uncertainty in Artificial Intelligence [UAI]

work page 2020
[45]

Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., Desmaison, A., Kopf, A., Yang, E., DeVito, Z., Raison, M., Tejani, A., Chilamkurthy, S., Steiner, B., Fang, L., Bai, J., and Chintala, S. (2019). PyTorch: An Imperative Style, High-Performance Deep Learning Library . In Advances in Neur...

work page 2019
[46]

Perez-Ortiz, M., Rivasplata, O., Parrado-Hernandez, E., Guedj, B., and Shawe-Taylor, J. (2021). Progress in Self-Certified Neural Networks . In NeurIPS 2021 Workshop on Bayesian Deep Learning

work page 2021
[47]

P \'e rez-Ortiz, M., Rivasplata, O., Shawe-Taylor, J., and Szepesv \'a ri, C. (2021). Tighter Risk Certificates for Neural Networks . Journal of Machine Learning Research , 22(227):1--40

work page 2021
[48]

Reisizadeh, A., Farnia, F., Pedarsani, R., and Jadbabaie, A. (2020). Robust Federated Learning: The Case of Affine Distribution Shifts . In Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M., and Lin, H., editors, Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December...

work page 2020
[49]

M., and Szepesv \' a ri, C

Rivasplata, O., Tankasali, V. M., and Szepesv \' a ri, C. (2019). PAC-Bayes with Backprop . arXiv , abs/1908.07380

work page arXiv 2019
[50]

Rothfuss, J., Fortuin, V., Josifoski, M., and Krause, A. (2021). PACOH: Bayes-optimal meta-learning with PAC-guarantees . In International Conference on Machine Learning (ICML)

work page 2021
[51]

Rothfuss, J., Josifoski, M., Fortuin, V., and Krause, A. (2022). PAC-Bayesian Meta-Learning: From Theory to Practice . arXiv , abs/2211.07206

work page arXiv 2022
[52]

Sakhi, O., Chopin, N., and Alquier, P. (2022). PAC-Bayesian Offline Contextual Bandits With Guarantees . arXiv , abs/2210.13132. To appear in ICML 2023

work page arXiv 2022
[53]

Scott, J., Zakerinia, H., and Lampert, C. H. (2023). PeFLL: A Lifelong Learning Approach to Personalized Federated Learning . arXiv , abs/2306.05515

work page arXiv 2023
[54]

Sefidgaran, M., Chor, R., Zaidi, A., and Wan, Y. (2023). Federated Learning You May Communicate Less Often! arXiv , abs/2306.05862

work page arXiv 2023
[55]

Seldin, Y., Laviolette, F., Cesa-Bianchi, N., Shawe-Taylor, J., and Auer, P. (2012). PAC-Bayesian Inequalities for Martingales . IEEE Transactions on Information Theory

work page 2012
[56]

Seldin, Y., Laviolette, F., Shawe-Taylor, J., Peters, J., and Auer, P. (2011). PAC-Bayesian Analysis of Martingales and Multiarmed Bandits . arXiv , abs/1105.2416

work page internal anchor Pith review Pith/arXiv arXiv 2011
[57]

and Williamson, R

Shawe-Taylor, J. and Williamson, R. C. (1997). A PAC analysis of a Bayes estimator . In Proceedings of the 10th annual conference on Computational Learning Theory , pages 2--9. ACM

work page 1997
[58]

T., Yu, F

Suresh, A. T., Yu, F. X., Kumar, S., and McMahan, H. B. (2017). Distributed Mean Estimation with Limited Communication . In Precup, D. and Teh, Y. W., editors, Proceedings of the 34th International Conference on Machine Learning, ICML 2017, Sydney, NSW, Australia, 6-11 August 2017 , volume 70 of Proceedings of Machine Learning Research , pages 3329--3337. PMLR

work page 2017
[59]

Z., Yu, H., Cui, L., and Yang, Q

Tan, A. Z., Yu, H., Cui, L., and Yang, Q. (2022). Towards Personalized Federated Learning . IEEE Transactions on Neural Networks and Learning Systems , pages 1--17

work page 2022
[60]

Tolstikhin, I. O. and Seldin, Y. (2013). PAC-Bayes-Empirical-Bernstein Inequality . In Burges, C. J. C., Bottou, L., Ghahramani, Z., and Weinberger, K. Q., editors, Advances in Neural Information Processing Systems 26: 27th Annual Conference on Neural Information Processing Systems 2013. Proceedings of a meeting held December 5-8, 2013, Lake Tahoe, Nevada...

work page 2013
[61]

Tsipras, D., Santurkar, S., Engstrom, L., Turner, A., and Madry, A. (2019). Robustness May Be at Odds with Accuracy

work page 2019
[62]

V., Mansfield, P

Vedadi, E., Dillon, J. V., Mansfield, P. A., Singhal, K., Afkanpour, A., and Morningstar, W. R. (2023). Federated Variational Inference: Towards Improved Personalization and Generalization . arXiv , abs/2305.13672

work page arXiv 2023
[63]

Yadan, O. (2019). Hydra - A framework for elegantly configuring complex applications . Github

work page 2019
[64]

Yagli, S., Dytso, A., and Vincent Poor, H. (2020). Information-Theoretic Bounds on the Generalization Error and Privacy Leakage in Federated Learning . In 2020 IEEE 21st International Workshop on Signal Processing Advances in Wireless Communications (SPAWC) , pages 1--5

work page 2020
[65]

B., Jette, M

Yoo, A. B., Jette, M. A., and Grondona, M. (2003). Slurm: Simple linux utility for resource management. In Feitelson, D., Rudolph, L., and Schwiegelshohn, U., editors, Job Scheduling Strategies for Parallel Processing , pages 44--60, Berlin, Heidelberg. Springer Berlin Heidelberg

work page 2003
[66]

Yurochkin, M., Agarwal, M., Ghosh, S., Greenewald, K., Hoang, N., and Khazaeni, Y. (2019). Bayesian Nonparametric Federated Learning of Neural Networks . In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning , volume 97 of Proceedings of Machine Learning Research , pages 7252--7261. PMLR

work page 2019
[67]

Zantedeschi, V., Viallard, P., Morvant, E., Emonet, R., Habrard, A., Germain, P., and Guedj, B. (2021). Learning Stochastic Majority Votes by Minimizing a PAC-Bayes Generalization Bound . In Conference on Neural Information Processing Systems (NeurIPS)

work page 2021
[68]

Zhang, C., Xie, Y., Bai, H., Yu, B., Li, W., and Gao, Y. (2021). A survey on federated learning. Knowledge-Based Systems , 216:106775

work page 2021
[69]

E., and Jordan, M

Zhang, H., Yu, Y., Jiao, J., Xing, E., Ghaoui, L. E., and Jordan, M. (2019). Theoretically Principled Trade-off between Robustness and Accuracy . In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning , volume 97 of Proceedings of Machine Learning Research , pages 7472--7482. PMLR

work page 2019
[70]

Zhang, L., Lei, X., Shi, Y., Huang, H., and Chen, C. (2023a). Federated Learning for IoT Devices With Domain Generalization . IEEE Internet Things J. , 10(11):9622--9633

work page
[71]

Zhang, X., Huang, A., Fan, L., Chen, K., and Yang, Q. (2023b). Probably Approximately Correct Federated Learning

work page
[72]

Zhang, X., Li, Y., Li, W., Guo, K., and Shao, Y. (2022). Personalized Federated Learning via Variational B ayesian Inference . In Chaudhuri, K., Jegelka, S., Song, L., Szepesvari, C., Niu, G., and Sabato, S., editors, Proceedings of the 39th International Conference on Machine Learning , volume 162 of Proceedings of Machine Learning Research , pages 26293...

work page 2022
[73]

Zhuang, F., Qi, Z., Duan, K., Xi, D., Zhu, Y., Zhu, H., Xiong, H., and He, Q. (2021). A Comprehensive Survey on Transfer Learning . Proc. IEEE , 109(1):43--76

work page 2021

[1] [1]

Achituve, I., Shamsian, A., Navon, A., Chechik, G., and Fetaya, E. (2021). Personalized Federated Learning With Gaussian Processes . In Ranzato, M., Beygelzimer, A., Dauphin, Y. N., Liang, P., and Vaughan, J. W., editors, Advances in Neural Information Processing Systems 34: Annual Conference on Neural Information Processing Systems 2021, NeurIPS 2021, De...

work page 2021

[2] [2]

T., Yu, F

Agarwal, N., Suresh, A. T., Yu, F. X., Kumar, S., and McMahan, B. (2018). cpSGD: Communication-efficient and differentially-private distributed SGD . In Bengio, S., Wallach, H. M., Larochelle, H., Grauman, K., Cesa - Bianchi, N., and Garnett, R., editors, Advances in Neural Information Processing Systems 31: Annual Conference on Neural Information Process...

work page 2018

[3] [3]

Alquier, P. (2021). User-friendly introduction to PAC-Bayes bounds . arXiv , abs/2110.11216

work page arXiv 2021

[4] [4]

Alquier, P., Ridgway, J., and Chopin, N. (2016). On the properties of variational approximations of G ibbs posteriors . Journal of Machine Learning Research

work page 2016

[5] [5]

and Meir, R

Amit, R. and Meir, R. (2018). Meta-Learning by Adjusting Priors Based on Extended PAC-Bayes Theory . In International Conference on Machine Learning ( ICML )

work page 2018

[6] [6]

Flower: A Friendly Federated Learning Research Framework

Beutel, D. J., Topal, T., Mathur, A., Qiu, X., Fernandez-Marques, J., Gao, Y., Sani, L., Kwing, H. L., Parcollet, T., Gusmão, P. P. d., and Lane, N. D. (2020). Flower: A Friendly Federated Learning Research Framework . arXiv preprint arXiv:2007.14390

work page internal anchor Pith review Pith/arXiv arXiv 2020

[7] [7]

and Guedj, B

Biggs, F. and Guedj, B. (2021). Differentiable PAC-Bayes objectives with partially aggregated neural networks. Entropy , 23(10)

work page 2021

[8] [8]

and Guedj, B

Biggs, F. and Guedj, B. (2022). Non-vacuous generalisation bounds for shallow neural networks. In Proceedings of the 39th International Conference on Machine Learning [ICML] , volume 162 of Proceedings of Machine Learning Research , pages 1963--1981. PMLR

work page 2022

[9] [9]

Biggs, F., Zantedeschi, V., and Guedj, B. (2022). On margins and generalisation for voting classifiers. In NeurIPS

work page 2022

[10] [10]

B., Patel, S., Ramage, D., Segal, A., and Seth, K

Bonawitz, K., Ivanov, V., Kreuter, B., Marcedone, A., McMahan, H. B., Patel, S., Ramage, D., Segal, A., and Seth, K. (2017). Practical secure aggregation for privacy-preserving machine learning. In Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security , pages 1175--1191

work page 2017

[11] [11]

Boucheron, S., Lugosi, G., and Massart, P. (2013). Concentration Inequalities - A Nonasymptotic Theory of Independence . Oxford University Press

work page 2013

[12] [12]

Catoni, O. (2007). PAC- Bayesian supervised classification: the thermodynamics of statistical learning . Institute of Mathematical Statistics

work page 2007

[13] [13]

and Chao, W

Chen, H. and Chao, W. (2021). FedBE: Making Bayesian Model Ensemble Applicable to Federated Learning . In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021 . OpenReview.net

work page 2021

[14] [14]

Ch\'erief-Abdellatif, B.-E., Shi, Y., Doucet, A., and Guedj, B. (2022). On PAC-Bayesian reconstruction guarantees for VAEs . In Proceedings of The 25th International Conference on Artificial Intelligence and Statistics [AISTATS] , volume 151 of Proceedings of Machine Learning Research , pages 3066--3079. PMLR

work page 2022

[15] [15]

Chugg, B., Wang, H., and Ramdas, A. (2023). A unified recipe for deriving (time-uniform) PAC-B ayes bounds . arXiv , abs/2302.03421

work page arXiv 2023

[16] [16]

Ding, N., Chen, X., Levinboim, T., Goodman, S., and Soricut, R. (2021). Bridging the Gap Between Practice and PAC-B ayes Theory in Few-Shot Meta-Learning . In Conference on Neural Information Processing Systems (NeurIPS)

work page 2021

[17] [17]

Duchi, J. (2007). Derivations for linear algebra and optimization . Berkeley, California , 3(1):2325--5870

work page 2007

[18] [18]

Dziugaite, G. K. and Roy, D. (2017). Computing Nonvacuous Generalization Bounds for Deep (Stochastic) Neural Networks with Many More Parameters than Training Data . In Conference on Uncertainty in Artificial Intelligence ( UAI )

work page 2017

[19] [19]

Fard, M. M. and Pineau, J. (2010). PAC-Bayesian model selection for reinforcement learning. In Conference on Neural Information Processing Systems (NeurIPS)

work page 2010

[20] [20]

and Majumdar, A

Farid, A. and Majumdar, A. (2021). Generalization Bounds for Meta-Learning via PAC-B ayes and Uniform Stability . In Conference on Neural Information Processing Systems (NeurIPS)

work page 2021

[21] [21]

Guedj, B. (2019). A Primer on PAC-Bayesian Learning . In Proceedings of the second congress of the French Mathematical Society , volume 33

work page 2019

[22] [22]

and Guedj, B

Haddouche, M. and Guedj, B. (2022). Online PAC-Bayes Learning . In Conference on Neural Information Processing Systems (NeurIPS)

work page 2022

[23] [23]

and Guedj, B

Haddouche, M. and Guedj, B. (2023). PAC-Bayes Generalisation Bounds for Heavy-Tailed Losses through Supermartingales . Transactions on Machine Learning Research

work page 2023

[24] [24]

Hardt, M., Price, E., and Srebro, N. (2016). Equality of Opportunity in Supervised Learning . In Lee, D. D., Sugiyama, M., von Luxburg, U., Guyon, I., and Garnett, R., editors, Advances in Neural Information Processing Systems 29: Annual Conference on Neural Information Processing Systems 2016, December 5-10, 2016, Barcelona, Spain , pages 3315--3323

work page 2016

[25] [25]

Hellström, F., Durisi, G., Guedj, B., and Raginsky, M. (2023). Generalization Bounds: Perspectives from Information Theory and PAC-Bayes . arXiv

work page 2023

[26] [26]

Jang, K., Jun, K.-S., Kuzborskij, I., and Orabona, F. (2023). Tighter PAC-Bayes Bounds Through Coin-Betting . arXiv , abs/2302.05829

work page arXiv 2023

[27] [27]

B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A

Kairouz, P., McMahan, H. B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A. N., Bonawitz, K., Charles, Z., Cormode, G., Cummings, R., D'Oliveira, R. G. L., Eichner, H., Rouayheb, S. E., Evans, D., Gardner, J., Garrett, Z., Gasc \'o n, A., Ghazi, B., Gibbons, P. B., Gruteser, M., Harchaoui, Z., He, C., He, L., Huo, Z., Hutchinson, B., Hsu, J., Jaggi, M., J...

work page 2021

[28] [28]

and Hospedales, T

Kim, M. and Hospedales, T. M. (2023). FedHB: Hierarchical Bayesian Federated Learning . arXiv , abs/2305.04979

work page arXiv 2023

[29] [29]

Kingma, D. P. and Welling, M. (2014). Auto-encoding variational bayes. In Bengio, Y. and LeCun, Y., editors, 2nd International Conference on Learning Representations, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings

work page 2014

[30] [30]

Federated Optimization: Distributed Machine Learning for On-Device Intelligence

Kone c n \'y , J., McMahan, H. B., Ramage, D., and Richt \' a rik, P. (2016a). Federated Optimization: Distributed Machine Learning for On-Device Intelligence . arXiv , abs/1610.02527

work page internal anchor Pith review Pith/arXiv arXiv

[31] [31]

Federated Learning: Strategies for Improving Communication Efficiency

Kone c n \'y , J., McMahan, H. B., Yu, F. X., Richt \' a rik, P., Suresh, A. T., and Bacon, D. (2016b). Federated Learning: Strategies for Improving Communication Efficiency . arXiv , abs/1610.05492

work page internal anchor Pith review Pith/arXiv arXiv

[32] [32]

Kotelevskii, N., Vono, M., Durmus, A., and Moulines, E. (2022). FedPop: A Bayesian Approach for Personalised Federated Learning . In NeurIPS

work page 2022

[33] [33]

and Szepesvári, C

Kuzborskij, I. and Szepesvári, C. (2019). Efron- S tein PAC - B ayesian I nequalities. arXiv:1909.01931

work page arXiv 2019

[34] [34]

Letarte, G., Germain, P., Guedj, B., and Laviolette, F. (2019). Dichotomize and generalize: PAC-Bayesian binary activated deep neural networks. In Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems [NeurIPS] 2019, 8-14 December 2019, Vancouver, BC, Canada , pages 6869--6879

work page 2019

[35] [35]

Li, L., Guedj, B., and Loustau, S. (2018). A quasi- Bayesian perspective to online clustering. Electron. J. Statist. , 12(2):3071--3113

work page 2018

[36] [36]

Mammen, P. M. (2021). Federated Learning: Opportunities and Challenges

work page 2021

[37] [37]

Maurer, A. (2004). A note on the PAC-Bayesian theorem . arXiv , cs/0411099

work page internal anchor Pith review Pith/arXiv arXiv 2004

[38] [38]

McAllester, D. A. (1998). Some PAC-Bayesian theorems . In Proceedings of the eleventh annual conference on Computational Learning Theory , pages 230--234. ACM

work page 1998

[39] [39]

McAllester, D. A. (2003). PAC-Bayesian stochastic model selection . Machine Learning , 51(1):5--21

work page 2003

[40] [40]

McMahan, B., Moore, E., Ramage, D., Hampson, S., and Arcas, B. A. y. (2017a). Communication-Efficient Learning of Deep Networks from Decentralized Data . In Singh, A. and Zhu, J., editors, Proceedings of the 20th International Conference on Artificial Intelligence and Statistics , volume 54 of Proceedings of Machine Learning Research , pages 1273--1282. PMLR

work page

[41] [41]

B., Moore, E., Ramage, D., Hampson, S., and y Arcas, B

McMahan, H. B., Moore, E., Ramage, D., Hampson, S., and y Arcas, B. A. (2017b). Communication-Efficient Learning of Deep Networks from Decentralized Data

work page

[42] [42]

Mhammedi, Z., Gr \" u nwald, P., and Guedj, B. (2019). PAC-Bayes un-expected Bernstein inequality. In Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems [NeurIPS] 2019, 8-14 December 2019, Vancouver, BC, Canada , pages 12180--12191

work page 2019

[43] [43]

Mohri, M., Sivek, G., and Suresh, A. T. (2019). Agnostic Federated Learning . In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning , volume 97 of Proceedings of Machine Learning Research , pages 4615--4625. PMLR

work page 2019

[44] [44]

Nozawa, K., Germain, P., and Guedj, B. (2020). PAC-Bayesian contrastive unsupervised representation learning. In Conference on Uncertainty in Artificial Intelligence [UAI]

work page 2020

[45] [45]

Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., Desmaison, A., Kopf, A., Yang, E., DeVito, Z., Raison, M., Tejani, A., Chilamkurthy, S., Steiner, B., Fang, L., Bai, J., and Chintala, S. (2019). PyTorch: An Imperative Style, High-Performance Deep Learning Library . In Advances in Neur...

work page 2019

[46] [46]

Perez-Ortiz, M., Rivasplata, O., Parrado-Hernandez, E., Guedj, B., and Shawe-Taylor, J. (2021). Progress in Self-Certified Neural Networks . In NeurIPS 2021 Workshop on Bayesian Deep Learning

work page 2021

[47] [47]

P \'e rez-Ortiz, M., Rivasplata, O., Shawe-Taylor, J., and Szepesv \'a ri, C. (2021). Tighter Risk Certificates for Neural Networks . Journal of Machine Learning Research , 22(227):1--40

work page 2021

[48] [48]

Reisizadeh, A., Farnia, F., Pedarsani, R., and Jadbabaie, A. (2020). Robust Federated Learning: The Case of Affine Distribution Shifts . In Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M., and Lin, H., editors, Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December...

work page 2020

[49] [49]

M., and Szepesv \' a ri, C

Rivasplata, O., Tankasali, V. M., and Szepesv \' a ri, C. (2019). PAC-Bayes with Backprop . arXiv , abs/1908.07380

work page arXiv 2019

[50] [50]

Rothfuss, J., Fortuin, V., Josifoski, M., and Krause, A. (2021). PACOH: Bayes-optimal meta-learning with PAC-guarantees . In International Conference on Machine Learning (ICML)

work page 2021

[51] [51]

Rothfuss, J., Josifoski, M., Fortuin, V., and Krause, A. (2022). PAC-Bayesian Meta-Learning: From Theory to Practice . arXiv , abs/2211.07206

work page arXiv 2022

[52] [52]

Sakhi, O., Chopin, N., and Alquier, P. (2022). PAC-Bayesian Offline Contextual Bandits With Guarantees . arXiv , abs/2210.13132. To appear in ICML 2023

work page arXiv 2022

[53] [53]

Scott, J., Zakerinia, H., and Lampert, C. H. (2023). PeFLL: A Lifelong Learning Approach to Personalized Federated Learning . arXiv , abs/2306.05515

work page arXiv 2023

[54] [54]

Sefidgaran, M., Chor, R., Zaidi, A., and Wan, Y. (2023). Federated Learning You May Communicate Less Often! arXiv , abs/2306.05862

work page arXiv 2023

[55] [55]

Seldin, Y., Laviolette, F., Cesa-Bianchi, N., Shawe-Taylor, J., and Auer, P. (2012). PAC-Bayesian Inequalities for Martingales . IEEE Transactions on Information Theory

work page 2012

[56] [56]

Seldin, Y., Laviolette, F., Shawe-Taylor, J., Peters, J., and Auer, P. (2011). PAC-Bayesian Analysis of Martingales and Multiarmed Bandits . arXiv , abs/1105.2416

work page internal anchor Pith review Pith/arXiv arXiv 2011

[57] [57]

and Williamson, R

Shawe-Taylor, J. and Williamson, R. C. (1997). A PAC analysis of a Bayes estimator . In Proceedings of the 10th annual conference on Computational Learning Theory , pages 2--9. ACM

work page 1997

[58] [58]

T., Yu, F

Suresh, A. T., Yu, F. X., Kumar, S., and McMahan, H. B. (2017). Distributed Mean Estimation with Limited Communication . In Precup, D. and Teh, Y. W., editors, Proceedings of the 34th International Conference on Machine Learning, ICML 2017, Sydney, NSW, Australia, 6-11 August 2017 , volume 70 of Proceedings of Machine Learning Research , pages 3329--3337. PMLR

work page 2017

[59] [59]

Z., Yu, H., Cui, L., and Yang, Q

Tan, A. Z., Yu, H., Cui, L., and Yang, Q. (2022). Towards Personalized Federated Learning . IEEE Transactions on Neural Networks and Learning Systems , pages 1--17

work page 2022

[60] [60]

Tolstikhin, I. O. and Seldin, Y. (2013). PAC-Bayes-Empirical-Bernstein Inequality . In Burges, C. J. C., Bottou, L., Ghahramani, Z., and Weinberger, K. Q., editors, Advances in Neural Information Processing Systems 26: 27th Annual Conference on Neural Information Processing Systems 2013. Proceedings of a meeting held December 5-8, 2013, Lake Tahoe, Nevada...

work page 2013

[61] [61]

Tsipras, D., Santurkar, S., Engstrom, L., Turner, A., and Madry, A. (2019). Robustness May Be at Odds with Accuracy

work page 2019

[62] [62]

V., Mansfield, P

Vedadi, E., Dillon, J. V., Mansfield, P. A., Singhal, K., Afkanpour, A., and Morningstar, W. R. (2023). Federated Variational Inference: Towards Improved Personalization and Generalization . arXiv , abs/2305.13672

work page arXiv 2023

[63] [63]

Yadan, O. (2019). Hydra - A framework for elegantly configuring complex applications . Github

work page 2019

[64] [64]

Yagli, S., Dytso, A., and Vincent Poor, H. (2020). Information-Theoretic Bounds on the Generalization Error and Privacy Leakage in Federated Learning . In 2020 IEEE 21st International Workshop on Signal Processing Advances in Wireless Communications (SPAWC) , pages 1--5

work page 2020

[65] [65]

B., Jette, M

Yoo, A. B., Jette, M. A., and Grondona, M. (2003). Slurm: Simple linux utility for resource management. In Feitelson, D., Rudolph, L., and Schwiegelshohn, U., editors, Job Scheduling Strategies for Parallel Processing , pages 44--60, Berlin, Heidelberg. Springer Berlin Heidelberg

work page 2003

[66] [66]

Yurochkin, M., Agarwal, M., Ghosh, S., Greenewald, K., Hoang, N., and Khazaeni, Y. (2019). Bayesian Nonparametric Federated Learning of Neural Networks . In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning , volume 97 of Proceedings of Machine Learning Research , pages 7252--7261. PMLR

work page 2019

[67] [67]

Zantedeschi, V., Viallard, P., Morvant, E., Emonet, R., Habrard, A., Germain, P., and Guedj, B. (2021). Learning Stochastic Majority Votes by Minimizing a PAC-Bayes Generalization Bound . In Conference on Neural Information Processing Systems (NeurIPS)

work page 2021

[68] [68]

Zhang, C., Xie, Y., Bai, H., Yu, B., Li, W., and Gao, Y. (2021). A survey on federated learning. Knowledge-Based Systems , 216:106775

work page 2021

[69] [69]

E., and Jordan, M

Zhang, H., Yu, Y., Jiao, J., Xing, E., Ghaoui, L. E., and Jordan, M. (2019). Theoretically Principled Trade-off between Robustness and Accuracy . In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning , volume 97 of Proceedings of Machine Learning Research , pages 7472--7482. PMLR

work page 2019

[70] [70]

Zhang, L., Lei, X., Shi, Y., Huang, H., and Chen, C. (2023a). Federated Learning for IoT Devices With Domain Generalization . IEEE Internet Things J. , 10(11):9622--9633

work page

[71] [71]

Zhang, X., Huang, A., Fan, L., Chen, K., and Yang, Q. (2023b). Probably Approximately Correct Federated Learning

work page

[72] [72]

Zhang, X., Li, Y., Li, W., Guo, K., and Shao, Y. (2022). Personalized Federated Learning via Variational B ayesian Inference . In Chaudhuri, K., Jegelka, S., Song, L., Szepesvari, C., Niu, G., and Sabato, S., editors, Proceedings of the 39th International Conference on Machine Learning , volume 162 of Proceedings of Machine Learning Research , pages 26293...

work page 2022

[73] [73]

Zhuang, F., Qi, Z., Duan, K., Xi, D., Zhu, Y., Zhu, H., Xiong, H., and He, Q. (2021). A Comprehensive Survey on Transfer Learning . Proc. IEEE , 109(1):43--76

work page 2021