Chisme: Heterogeneity-Aware Gossip Learning

Harikrishna Kuttivelil; Katia Obraczka

REVIEW 3 major objections 6 minor 29 references

Chisme: Heterogeneity-Aware Gossip Learning

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Chisme weights each received model by how closely its update direction matches the client's own, and claims this yields faster, fairer decentralized learning at the edge.

desk verdict A likeable gossip-learning paper with a normalization bug that undercuts its stated mechanism; the empirical claims may survive, but the story needs rework. read the letter →

arxiv 2505.09854 v3 pith:T7WSBMC3 submitted 2025-05-14 cs.LG cs.ETcs.MAcs.SI

Harikrishna Kuttivelil , Katia Obraczka This is my paper · ORCID

classification cs.LGcs.ETcs.MAcs.SI

keywords gossiplearningdecentralizedheterogeneousdatapersonalizedmodelscosinesimilaritymodelmergingedgecomputingnon-IID

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Chisme is a fully decentralized gossip-learning protocol that assigns each incoming neighbor model a merge weight derived from the cosine similarity between the neighbor's update direction and the client's own update direction. The paper argues that this scalar is a usable proxy for how similar the two clients' underlying data distributions are, so weighting merges by it lets each client cooperate broadly at first and then smoothly specialize toward its own data. If the claim holds, devices on lossy, infrastructure-less edge networks can converge faster and with more equitable client outcomes than vanilla gossip, federated, or clustering-based alternatives, without any extra communication. The reported experiments support this across image and weather-forecasting tasks, with the largest margins under weaker connectivity.

What carries the argument

The load-bearing object is the cosine-similarity data-affinity heuristic $\omega_{i,k}$. When client $i$ receives a message from client $k$, it measures its own change since its last saved state, $\Delta_{i,i} = \theta_i^{t_i} - \theta_i^{\tau_i}$, and the received model's change from the same saved state, $\Delta_{k,i} = \theta_k^* - \theta_i^{\tau_i}$. The cosine similarity $S$ of these two delta vectors is scaled to $S'=(S+1)/2\in[0,1]$, then turned into the heuristic $\omega_{i,k}=S'/(1+S')$. Chisme replaces the standard experience-only gossip merge weight $\alpha_k$ with the combined influence $\eta_k = \alpha_k\omega_{i,k}/((1-\alpha_k)(1-\omega_{i,k})+\alpha_k\omega_{i,k})$, so a received model with similar update direction gets a larger share of the merge while dissimilar ones are down-weighted. This mechanism carries the argument because it adds no extra message traffic and keeps per-client memory at about three model copies.

What would settle it

One deciding experiment is to simulate client pairs with known distribution overlap, compute the paper's $\omega_{i,k}$ while deliberately varying local epochs and merge histories, and check whether $\omega_{i,k}$ tracks held-out data affinity; if the rank correlation disappears once training budgets are mismatched, the heuristic is reweighting on asynchronous-training artifacts rather than data affinity.

Watch

Extended reading notes

Core claim

The paper claims that a single scalar computed from model parameter directions can differentiate collaboration in fully decentralized, asynchronous gossip learning, and that this differentiation benefits all clients. In experiments across image recognition and weather time-series prediction, clients running Chisme converge faster, reach lower mean loss, and show smaller standard deviation of loss across clients, with the margin growing when connectivity and reliability drop. The one exception the paper reports is the extreme label-swapped MNIST setting under favorable networks, where an incremental-clustering baseline reaches lower loss and lower disparity; Chisme still edges it out when the same setting is run over weaker networks.

Load-bearing premise

The load-bearing premise is that the angle between two clients' model-change vectors measures how similar their data distributions are, even when the two models have undergone unequal amounts of training and merging before the comparison.

Editorial extensions

If this is right

A gossip client can personalize its model without a separate finetuning or clustering stage, because merge weights adapt continuously as data affinities are discovered.
The mechanism works on top of the same messages vanilla gossip already exchanges, so bandwidth stays unchanged while memory stays near the cost of three model copies.
Under episodic or lossy links, the expected gain over non-differentiated baselines increases, making the approach a candidate for remote-sensor and disaster-response settings.
Equity across clients improves in the reported setups: the standard deviation of client loss falls, so no single client is left with a poor model under heterogeneity.

Reading between the lines

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

An extension the paper does not test: if the cosine proxy is stable, the same $\omega_{i,k}$ values could be logged as a privacy-preserving signal of which clients share data characteristics, since only parameter vectors are exchanged.
A cheap robustness check not reported in the paper would normalize $\Delta_{i,i}$ and $\Delta_{k,i}$ by the number of local epochs since the saved state before computing $S$; the paper's central assumption predicts this changes merge rankings, while a competing explanation predicts it leaves performance roughly unchanged.
The smooth broad-to-selective collaboration transition implies a tunable continuum between global averaging and pure local training; adding a sigmoid or temperature on $\omega$ would let applications trade generalization against personalization, which the paper leaves as future work.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick take: Chisme is the first fully asynchronous gossip scheme I know that uses cosine similarity between model deltas to weight merges. That is a real, modest novelty. The writing is clear, and the evaluation spans two modalities and two network regimes. The paper deserves a serious referee. But the core mechanism as written does not match the abstract, and the empirical section is weaker than the claims.

What is new: unlike CFL/FLIC, there is no central clustering; unlike [23], no synchronous rounds; unlike [24], no validation set or extra loss computation. The update rule only needs the local snapshot and the incoming model, so memory stays constant at 3p. That is a useful trick for edge/IoT.

Soft spots, in order:

1. Eq. 14 caps omega at 0.5 (since S' is in [0,1]), so Eq. 15 always gives eta <= alpha. The abstract says Chisme "facilitates stronger merging influence between more similar clients"; relative to less similar clients that is true, since eta increases with S', but relative to the standard experience-based influence alpha, Chisme never boosts collaboration, it only damps it. If the intended claim is "stronger than vanilla GL," the formula does not deliver that. If the intended claim is only "more selective," then the wording in the abstract and Section III-D is misleading and should be changed. The stress-test algebra checks out. This is fixable by redefining omega (for instance omega = 2S'/(1+S') or simply S') and re-running the experiments.

2. The affinity signal is contaminated. Delta_i,i is measured from the pre-training snapshot, but after that snapshot the active model may have already absorbed several received models. So S' reflects prior merges, not cleanly P_i vs P_k. The paper does not test whether omega actually tracks true data affinity; an ablation with oracle or shuffled affinities would help a lot.

3. The experiments have no error bars or statistical tests, and no code is released. "Over multiple seeds" is not enough to support "outperforms in almost every case." Also, the comparison set omits the most relevant personalization baselines, especially Dinani et al. [24], which they cite but do not benchmark.

The central idea is plausible and the paper is readable. The flaws are not fatal, but they are load-bearing for the story. A referee should ask for a corrected normalization, a contamination analysis, and proper statistics before acceptance. Citation pattern looks fine to me.

Referee Report

3 major / 6 minor

Summary. The paper presents Chisme, a fully decentralized gossip learning algorithm that uses cosine similarity between local and received model update vectors as a heuristic for data-distribution affinity, and uses this heuristic to modulate the merge weight when incorporating received models. The authors evaluate Chisme on FEMNIST, label-swapped MNIST, and Irish weather data under two network configurations, and report that Chisme generally achieves faster convergence, lower average loss, and lower cross-client loss disparity compared with FL, DFL, GL, FLIC, and their own CosSimDFL baseline. The paper claims that Chisme realizes a smooth transition from broad to selective collaboration, and that its advantages are more pronounced under less favorable network conditions.

Significance. If the reported results hold, the paper would make a useful contribution to decentralized personalized learning: it provides a fully asynchronous, memory-bounded method (constant O(p) memory) that does not require central clustering or synchronous rounds, and the evaluation spans real and synthetic heterogeneity. The paper also gives a clear analysis of the memory footprint and positions the work relative to clustered federated learning. However, the central mechanism as described is internally inconsistent: the normalization in Eq. 14 ensures the affinity weight never exceeds 0.5, so the combined influence in Eq. 15 is always less than or equal to the base experience-based weight. Consequently, the abstract's claim of 'stronger merging influence' for similar clients is not supported by the algorithm's algebra, and the empirical gains could stem from a generic damping of the merge step rather than affinity-aware weighting. The affinity estimate itself is also not validated against ground-truth data similarity, leaving the mechanism's correctness unestablished. These issues are load-bearing for the paper's claimed contribution.

major comments (3)

[III-D, Eqs. (14)-(15)] Since S' is in [0,1], Eq. 14 gives ω = S'/(1+S') ≤ 0.5. For α in [0,1], Eq. 15 yields η = αω / ((1-α)(1-ω)+αω). Rearranging, η ≤ α whenever ω ≤ 0.5, with equality only when S'=1. Therefore Chisme never produces a stronger merging influence for similar clients than the standard experience-based weight; it always damps the incoming update (or at best leaves it unchanged for perfectly aligned deltas). This contradicts the abstract and Section I's claim that the method 'facilitates stronger merging influence between clients with more similar model learning progressions.' The authors should either correct the claim, change the heuristic so that similar clients can receive η > α, or explicitly reframe Chisme as an affinity-dependent damping mechanism and provide evidence that the damping, rather than the affinity discrimination, is the source of the reported improvements.
[III-B and III-D, Eqs. (5)-(6), (14)] The affinity heuristic is computed as the cosine similarity between Δ_{i,i} = θ^{t_i}_i - θ^{τ_i}_i and Δ_{k,i} = θ^*_k - θ^{τ_i}_i, where θ^{τ_i}_i is the last pre-training snapshot. However, θ^{τ_i}_i is taken after all previous merges at client i, so Δ_{i,i} contains the residual of prior remote updates already incorporated into the local model. Similarly, θ^*_k has an unknown merge history. Thus S' does not isolate P_i versus P_k; it is a function of the entire past exchange trajectory. The paper does not test whether ω tracks true data-distribution affinity, e.g., by comparing cosine similarities between clients with known label overlap or by ablating with random or constant weights. Without such validation, the reported performance gains are equally consistent with a global reduction in collaboration intensity.
[IV-B and Figures 2-4] The central empirical claim is that 'in almost every case, Chisme outperforms all other approaches,' but the paper provides no numerical comparisons with variance or statistical significance. The figures plot mean curves over multiple seeds without confidence intervals or error bars, and no table reports final losses or standard deviations. In the favorable MNIST configuration (Figure 3a), FLIC clearly outperforms Chisme, which undercuts the blanket wording. The authors should quantify the comparisons (e.g., mean ± std over seeds, win/tie/loss counts, or paired tests) and specify the number of seeds and how randomness is controlled.

minor comments (6)

[III-D, Eq. (15)] The denominator in Eq. (15) contains 'αω', which is ambiguous; it should read 'α_k ω_{i,k}' (or 'α ω_{i,k}') for consistency with the other indexed quantities.
[Abstract and Section I] The phrase 'stronger merging influence between clients with more similar model learning progressions' is contradicted by the algebra of Eqs. (14)-(15); please rephrase to describe the actual effect of the affinity heuristic on the merge weight.
[III-A, Eqs. (3)-(4)] The indexing in the local experience map M is unclear: Eq. (3) writes 'M_k^x' but the intended meaning appears to be 'M_i^k' (client i's stored experience for neighbor k). Please clean up the notation.
[IV-A] The definition of connectivity C_N = 0 as 'a minimally connected ring topology' is not immediately consistent with the Watts-Strogatz model; please clarify how intermediate C_N values are mapped to topology generation.
[IV-B] There is a typo: 'analagous' should be 'analogous'.
[III-D, Note on Memory & Convergence] The convergence argument is only a heuristic analogy to GL and FL; please state the precise assumptions under which convergence is expected, or label this as an empirical observation rather than a justification.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the cosine-similarity heuristic is externally sourced, parameter-free, and benchmarked against independent baselines.

full rationale

Chisme does not derive a fitted prediction from its inputs. Its core mechanism is a fixed, parameter-free cosine-similarity heuristic applied to model deltas (Eqs. 5-7), adapted from external prior work (CFL [21], FLIC [22], and personalized DFL [23]) and openly labeled a heuristic. The merge coefficients in Eqs. 14-16 are algebraic functions of that heuristic; they are not fitted to the benchmark losses, and the benchmark comparisons (FEMNIST, label-swapped MNIST, Irish weather) are independent external datasets with standard baselines (FL, DFL, GL, FLIC, CosSimDFL). There are no self-citations, no imported uniqueness theorems, and no renamed known result: the algorithm combines known components in a new asynchronous gossip setting. The paper's own convergence note is an assumption rather than a derived result, and Eq. 14's normalization and Eq. 15's damping property (eta <= alpha whenever S' <= 1) are correctness or bug concerns about the strength of the claimed effect, not circular reasoning. Therefore no circular step can be exhibited under the required evidence standard.

Assumptions & free parameters 2 free parameters · 4 assumptions · 1 invented entities

The central claim depends on a set of domain assumptions about initialization, topology, and the reliability of cosine similarity as an affinity signal, plus a hand-chosen normalization that is not derived. No parameters are fitted to external data, but the normalization constants in Eqs. 7 and 14 are chosen by hand and determine the algorithm's behavior.

free parameters (2)

Scaled-cosine offset (1/2 in Eq. 7) = 0.5
Hand-chosen transform to map cosine similarity [-1,1] to [0,1]; the central merge weight depends on this scaling.
Normalization denominator in Eq. 14 (1 + S') = 1
Hand-chosen normalization that caps the affinity heuristic at 0.5, preventing any boost for similar clients. This is not derived and is inconsistent with the abstract's claim of stronger merging influence.

assumptions (4)

domain assumption All clients are initialized with identical model parameters theta^{t=0}.
Stated in Section III-D INITIALIZATION (Eq. 9); the cosine similarity argument relies on a common baseline state.
domain assumption The underlying network topology is static and communication cost between neighbors is constant.
Stated in Section III system model; the algorithm's guarantees are scoped to this setting.
domain assumption Models trained on similar data distributions move in similar directions in parameter space, so cosine similarity of deltas indicates data affinity.
This is the core premise of Section III-B and is not proven; it is the load-bearing assumption behind the heuristic.
domain assumption Convergence of Chisme follows from convergence of standard GL and FL when collaboration is strong and from on-device learning when collaboration is low.
Section III-D 'Note on Memory & Convergence' states 'we can assume that it will converge'; no proof is provided.

invented entities (1)

omega_i,k (normalized similarity-based affinity heuristic)
purpose: Determines the weight of a received model in the merge step (Eqs. 14-15).
A hand-defined function of model deltas; the paper provides no external benchmark or theoretical derivation showing it measures data affinity. Its behavior is also inconsistent with the stated goal because it never exceeds 0.5.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Chisme: Heterogeneity-Aware Gossip Learning." pith.science (2026). https://pith.science/paper/T7WSBMC3

@misc{pith2026250509854,
  author       = {Pith},
  title        = {Pith review of: Chisme: Heterogeneity-Aware Gossip Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T7WSBMC3}},
  note         = {Machine review of arXiv:2505.09854}
}

read the original abstract

As end-user device capability increases and demand for intelligent services at the Internet's edge rises, distributed learning has emerged as a key enabling technology for the intelligent edge. Existing approaches like federated learning (FL) and decentralized FL (DFL) enable privacy-preserving distributed learning among clients, while gossip learning (GL) approaches have emerged to address the potential challenges in resource-constrained, connectivity-challenged infrastructure-less environments. However, most distributed learning approaches assume largely homogeneous data distributions and may not consider or exploit the heterogeneity of clients and their underlying data distributions. This paper introduces Chisme, a novel fully decentralized distributed learning algorithm designed to address the challenges of implementing robust intelligence in network edge contexts characterized by heterogeneous data distributions, episodic connectivity, and sparse network infrastructure or lack thereof. Chisme leverages the affinity between clients' underlying data distributions calculated from received model exchanges to inform how much influence received models have when merging into the local model. By doing so, it enables clients to strategically balance between broader collaboration to build more general knowledge and more selective collaboration to build specific knowledge. We evaluate Chisme against contemporary approaches using image recognition and time-series prediction scenarios while considering different network connectivity conditions, representative of real-world distributed intelligent systems running at the network's edge. Our experiments demonstrate that Chisme outperforms state-of-the-art edge intelligence approaches in almost every case -- clients using Chisme exhibit faster training convergence, lower final loss after training, and lower performance disparity between clients.

Figures

Figures reproduced from arXiv: 2505.09854 by the authors.

**Figure 2.** Loss mean and standard deviation (σ) across clients in the FEMNIST handwritten digit recognition scenario. differentiated FL experiments [21], [22]), in which 5 different data distribution groups swap different pairs of labels to introduce incongruence between clients’ datasets. The Irish weather station dataset was sourced from the Irish metereological service Met Eiranne ´ [26] for a single year across 25 weather… view at source ↗

**Figure 3.** Loss mean and standard deviation (σ) across clients in the artificially incongruent, label-swapped MNIST handwritten digit recognition scenario. B. Discussion We find that in almost every case, Chisme outperforms all other approaches, with clients’ joint model training converging faster to a lower loss while achieving lower performance disparity between clients. We observe its performance advantages are more pronou… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Loss mean and standard deviation (σ) across clients in the Irish Weather Station multi-feature weather prediction scenario. performance outcomes across non-IID clients. V. CONCLUSION In this paper, we introduced Chisme, a fully decentralized heterogeneity-aware approac…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 14 canonical work pages

[23]

Like attracts like: Personalized federated learning in decentralized edge computing,

Z. Ma, Y . Xu, H. Xu, J. Liu, and Y . Xue, “Like attracts like: Personalized federated learning in decentralized edge computing,”IEEE Transactions on Mobile Computing, vol. 23, no. 2, pp. 1080–1096, 2022

work page 2022
[24]

Gos- sip learning of personalized models for vehicle trajectory prediction,

M. A. Dinani, A. Holzer, H. Nguyen, M. A. Marsan, and G. Rizzo, “Gos- sip learning of personalized models for vehicle trajectory prediction,” in2021 IEEE Wireless Communications and Networking Conference Workshops (WCNCW). IEEE, 2021, pp. 1–7

work page 2021
[1]

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

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

2019
[2]

The de-democratization of ai: Deep learning and the compute divide in artificial intelligence research,

N. Ahmed and M. Wahed, “The de-democratization of ai: Deep learning and the compute divide in artificial intelligence research,”arXiv preprint arXiv:2010.15581, 2020

arXiv 2010
[3]

Overview of edge computing in the agricultural internet of things: Key technologies, applications, challenges,

X. Zhang, Z. Cao, and W. Dong, “Overview of edge computing in the agricultural internet of things: Key technologies, applications, challenges,”Ieee Access, vol. 8, pp. 141 748–141 761, 2020

2020
[4]

Smarter eco- cities and their leading-edge artificial intelligence of things solutions for environmental sustainability: A comprehensive systematic review,

S. E. Bibri, J. Krogstie, A. Kaboli, and A. Alahi, “Smarter eco- cities and their leading-edge artificial intelligence of things solutions for environmental sustainability: A comprehensive systematic review,” Environmental Science and Ecotechnology, vol. 19, p. 100330, 2024

work page 2024
[5]

Edge technologies for disaster management: A survey of social media and artificial intelligence integration,

M. Aboualola, K. Abualsaud, T. Khattab, N. Zorba, and H. S. Hassanein, “Edge technologies for disaster management: A survey of social media and artificial intelligence integration,”IEEE Access, 2023

work page 2023
[6]

Edge intelligence and internet of things in healthcare: A survey,

S. U. Amin and M. S. Hossain, “Edge intelligence and internet of things in healthcare: A survey,”Ieee Access, vol. 9, pp. 45–59, 2020

work page 2020

Show all 29 references

[7]

Mobile edge intelligence and computing for the internet of vehicles,

J. Zhang and K. B. Letaief, “Mobile edge intelligence and computing for the internet of vehicles,”Proceedings of the IEEE, vol. 108, no. 2, pp. 246–261, 2019

2019
[8]

Communication-efficient learning of deep networks from decentralized data,

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

2017
[9]

Gossip learning as a decen- tralized alternative to federated learning,

I. Heged ˝us, G. Danner, and M. Jelasity, “Gossip learning as a decen- tralized alternative to federated learning,” inDistributed Applications and Interoperable Systems: 19th IFIP WG 6.1 International Conference, DAIS 2019, Held as Part of the 14th International Federated Conf...

2019
[10]

Learning in multiagent systems,

S. Sen and G. Weiss, “Learning in multiagent systems,”Multiagent systems: A modern approach to distributed artificial intelligence, pp. 259–298, 1999

1999
[11]

Local & federated learning at the network edge for efficient predictive analytics,

N. Harth, C. Anagnostopoulos, H.-J. V oegel, and K. Kolomvatsos, “Local & federated learning at the network edge for efficient predictive analytics,”Future Generation Computer Systems, vol. 134, pp. 107–122, 2022

2022
[12]

Federated learning for smart healthcare: A survey,

D. C. Nguyen, Q.-V . Pham, P. N. Pathirana, M. Ding, A. Seneviratne, Z. Lin, O. Dobre, and W.-J. Hwang, “Federated learning for smart healthcare: A survey,”ACM Computing Surveys (Csur), vol. 55, no. 3, pp. 1–37, 2022

2022
[13]

Federated learning for internet of things: Recent advances, taxonomy, and open challenges,

L. U. Khan, W. Saad, Z. Han, E. Hossain, and C. S. Hong, “Federated learning for internet of things: Recent advances, taxonomy, and open challenges,”IEEE Communications Surveys & Tutorials, vol. 23, no. 3, pp. 1759–1799, 2021

2021
[14]

Federated learning for internet of things: A comprehensive survey,

D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li, and H. V . Poor, “Federated learning for internet of things: A comprehensive survey,”IEEE Communications Surveys & Tutorials, vol. 23, no. 3, pp. 1622–1658, 2021

2021
[15]

Fully decentralized federated learning,

A. Lalitha, S. Shekhar, T. Javidi, and F. Koushanfar, “Fully decentralized federated learning,” inThird workshop on bayesian deep learning (NeurIPS), vol. 2, 2018

2018
[16]

Decentralized federated learning: Balancing communication and computing costs,

W. Liu, L. Chen, and W. Zhang, “Decentralized federated learning: Balancing communication and computing costs,”IEEE Transactions on Signal and Information Processing over Networks, vol. 8, pp. 131–143, 2022

2022
[17]

Differentiating instruction in response to student readiness, interest, and learning profile in academically diverse classrooms: A review of literature,

C. A. Tomlinson, C. Brighton, H. Hertberg, C. M. Callahan, T. R. Moon, K. Brimijoin, L. A. Conover, and T. Reynolds, “Differentiating instruction in response to student readiness, interest, and learning profile in academically diverse classrooms: A review of literature,”Journa...

2003
[18]

Client-edge-cloud hierarchical federated learning,

L. Liu, J. Zhang, S. Song, and K. B. Letaief, “Client-edge-cloud hierarchical federated learning,” inICC 2020-2020 IEEE international conference on communications (ICC). IEEE, 2020, pp. 1–6

2020
[19]

Per- sonalized cross-silo federated learning on non-iid data,

Y . Huang, L. Chu, Z. Zhou, L. Wang, J. Liu, J. Pei, and Y . Zhang, “Per- sonalized cross-silo federated learning on non-iid data,” inProceedings of the AAAI conference on artificial intelligence, vol. 35, no. 9, 2021, pp. 7865–7873

2021
[20]

Federated learning with personalization layers,

M. G. Arivazhagan, V . Aggarwal, A. K. Singh, and S. Choud- hary, “Federated learning with personalization layers,”arXiv preprint arXiv:1912.00818, 2019

1912 arXiv
[21]

Clustered federated learning: Model-agnostic distributed multitask optimization under privacy con- straints,

F. Sattler, K.-R. M ¨uller, and W. Samek, “Clustered federated learning: Model-agnostic distributed multitask optimization under privacy con- straints,”IEEE transactions on neural networks and learning systems, vol. 32, no. 8, pp. 3710–3722, 2020

2020
[22]

Federated learning with incremental clustering for heterogeneous data,

F. E. Castellon, A. Mayoue, J.-H. Sublemontier, and C. Gouy-Pailler, “Federated learning with incremental clustering for heterogeneous data,” in2022 International Joint Conference on Neural Networks (IJCNN). IEEE, 2022, pp. 1–8

2022
[25]

Leaf: A benchmark for federated settings,

S. Caldas, S. M. K. Duddu, P. Wu, T. Li, J. Kone ˇcn`y, H. B. McMahan, V . Smith, and A. Talwalkar, “Leaf: A benchmark for federated settings,” arXiv preprint arXiv:1812.01097, 2018

2018 arXiv
[26]

Available data,

Met ´Eireann, “Available data,” https://www.met.ie/climate/ available-data, 2025

2025
[27]

Gradient-based learning applied to document recognition,

Y . LeCun, L. Bottou, Y . Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,”Proceedings of the IEEE, vol. 86, no. 11, pp. 2278–2324, 1998

1998
[28]

A multi-step wind power group forecasting seq2seq architecture with spatial–temporal feature fusion and numerical weather prediction cor- rection,

S. Xu, Y . Wang, X. Xu, G. Shi, Y . Zheng, H. Huang, and C. Hong, “A multi-step wind power group forecasting seq2seq architecture with spatial–temporal feature fusion and numerical weather prediction cor- rection,”Energy, vol. 291, p. 130352, 2024

2024
[29]

Collective dynamics of ‘small- world’networks,

D. J. Watts and S. H. Strogatz, “Collective dynamics of ‘small- world’networks,”nature, vol. 393, no. 6684, pp. 440–442, 1998

1998

Pith tools

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

[1] [23]

Like attracts like: Personalized federated learning in decentralized edge computing,

Z. Ma, Y . Xu, H. Xu, J. Liu, and Y . Xue, “Like attracts like: Personalized federated learning in decentralized edge computing,”IEEE Transactions on Mobile Computing, vol. 23, no. 2, pp. 1080–1096, 2022

work page 2022

[2] [24]

Gos- sip learning of personalized models for vehicle trajectory prediction,

M. A. Dinani, A. Holzer, H. Nguyen, M. A. Marsan, and G. Rizzo, “Gos- sip learning of personalized models for vehicle trajectory prediction,” in2021 IEEE Wireless Communications and Networking Conference Workshops (WCNCW). IEEE, 2021, pp. 1–7

work page 2021

[3] [1]

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

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

2019

[4] [2]

The de-democratization of ai: Deep learning and the compute divide in artificial intelligence research,

N. Ahmed and M. Wahed, “The de-democratization of ai: Deep learning and the compute divide in artificial intelligence research,”arXiv preprint arXiv:2010.15581, 2020

arXiv 2010

[5] [3]

Overview of edge computing in the agricultural internet of things: Key technologies, applications, challenges,

X. Zhang, Z. Cao, and W. Dong, “Overview of edge computing in the agricultural internet of things: Key technologies, applications, challenges,”Ieee Access, vol. 8, pp. 141 748–141 761, 2020

2020

[6] [4]

Smarter eco- cities and their leading-edge artificial intelligence of things solutions for environmental sustainability: A comprehensive systematic review,

S. E. Bibri, J. Krogstie, A. Kaboli, and A. Alahi, “Smarter eco- cities and their leading-edge artificial intelligence of things solutions for environmental sustainability: A comprehensive systematic review,” Environmental Science and Ecotechnology, vol. 19, p. 100330, 2024

work page 2024

[7] [5]

Edge technologies for disaster management: A survey of social media and artificial intelligence integration,

M. Aboualola, K. Abualsaud, T. Khattab, N. Zorba, and H. S. Hassanein, “Edge technologies for disaster management: A survey of social media and artificial intelligence integration,”IEEE Access, 2023

work page 2023

[8] [6]

Edge intelligence and internet of things in healthcare: A survey,

S. U. Amin and M. S. Hossain, “Edge intelligence and internet of things in healthcare: A survey,”Ieee Access, vol. 9, pp. 45–59, 2020

work page 2020