Pith. sign in

REVIEW 2 major objections 5 minor 58 references

Fedivertex: a Graph Dataset based on Decentralized Social Networks for Trustworthy Machine Learning

T0 review · 2 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper introduces Fedivertex, a public dataset of 182 temporal graphs from seven Fediverse platforms, and argues it is the first resource enabling reproducible machine-learning experiments on decentralized social networks, including a…

desk verdict Solid, reusable multi-platform Fediverse graph dataset; the 'defederation prediction' task is the one weak joint because weekly snapshots can't distinguish administrative blocks from outages or inactivity. read the letter →

arxiv 2505.20882 v1 pith:FKAFBOFU submitted 2025-05-27 cs.LG cs.SI

Marc Damie , Edwige Cyffers This is my paper
classification cs.LGcs.SI
keywords FediversedecentralizedsocialnetworkstemporalgraphdatasetdefederationpredictionmachinelearningbenchmarkingActivityPubcommunitydetection
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

The paper introduces Fedivertex, a publicly released dataset of 182 temporal graphs built from seven Fediverse platforms—Mastodon, Misskey, Pleroma, PeerTube, Friendica, Bookwyrm, and Lemmy—with 14 weekly snapshots per graph. Its central claim is that this is the first dataset enabling reproducible experiments on decentralized social-network topology at the instance level, which sidesteps the recommendation-driven link formation seen in for-profit platform graphs. The authors argue that the dataset captures dynamics standard graph collections miss, most notably defederation, the deliberate deletion of inter-instance links, and they formalize this as a new prediction task with baselines. A sympathetic reader should care because decentralized machine-learning algorithms depend on communication topology, and Fedivertex supplies real, timestamped topologies on which to benchmark them.

What carries the argument

The carrying mechanism is a weekly web-crawl pipeline that queries public instance-level endpoints from a community-maintained index of Fediverse servers and encodes the results as graphs. Three graph families carry the argument: federation graphs, where an edge means two instances have interacted; active-user graphs, where directed edge weights count follows from an instance's most recently active users; and Lemmy's intra-instance and cross-instance graphs, which encode where messages were posted and which threads were shared. The formal object for the new task is the pair of snapshots $G_t=(V_t,E_t)$ and $G_{t'}=(V_{t'},E_{t'})$; restricted to the common node set $V_c=V_t\cap V_{t'}$, the prediction target is the deleted edge set $E_t\setminus E_{t'}$ and the created edge set $E_{t'}\setminus E_t$. This snapshot-comparison construction is what turns the social process of defederation into a machine-learning benchmark.

What would settle it

A concrete test is to run a daily or event-driven crawl over the same instances for an interval overlapping the dataset's collection period and compare: if many edges that disappear within a day never appear in any weekly snapshot because they are re-established before the next weekly observation, the claim that Fedivertex supports defederation prediction is falsified. Instance availability logs, such as those kept by community indexers, could provide the same evidence without extra crawling.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that a reproducible, multi-platform, temporal graph resource for the Fediverse is feasible and useful. Fedivertex encodes instance-to-instance communication following the ActivityPub standard, yielding 13 graph types: federation graphs for six platforms, active-user graphs for Mastodon, Misskey, and Pleroma, a follow graph for PeerTube, and three Lemmy-specific graphs, each observed at 14 weekly timestamps. The paper shows that these graphs are small-world but differ from classic social-network graphs in degree distribution, and that they support three kinds of experiments: a differentially private random-walk decentralized learning benchmark, link creation and deletion prediction, and language-based community detection. It also introduces the defederation prediction task, in which the goal is to forecast which edges or nodes disappear between consecutive snapshots, and reports that topology-score baselines remain near random on federation graphs—a result the authors attribute to the speed of defederation relative to the weekly sampling.

Load-bearing premise

The load-bearing premise is that the weekly crawl interval resolves the deletion dynamics the defederation task targets; the paper itself notes that defederation is extremely quick and that weekly granularity is not enough to beat random baselines, so if most deletions occur and revert within a week, the defederation task becomes ill-posed even though the dataset would still support other benchmarks.

Editorial extensions

If this is right

  • Decentralized learning algorithms can be benchmarked on real Fediverse communication topologies rather than synthetic or for-profit social graphs, and the paper's reproduction of a random-walk private learning method shows that real topologies can slow convergence relative to synthetic ones.
  • Edge deletion becomes a measurable phenomenon on these graphs: across the 14 snapshots, some federation graphs grow while others are dominated by edge loss, and most deletions persist beyond two weeks, suggesting the process is not merely crawl noise.
  • The dataset makes temporal graph tasks beyond link creation testable, including defederation prediction and instance-dropout prediction, with concrete baselines and evaluation protocols.
  • The statistically distinct degree regime observed in Fedivertex graphs—smoother, more concave degree decay and higher normalized density than classic social-network graphs—provides a testbed for whether network-evolution models built on for-profit platforms transfer to volunteer-run decentralized networks.

Reading between the lines

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

  • Beyond the paper: a higher-frequency crawl, triggered by moderation events or run daily, could make the defederation task well-posed; the paper's own baselines are near random, and its stated explanation is that defederation is quicker than the weekly sampling.
  • Beyond the paper: recording explicit blocks and moderation decisions alongside silent federation drops would give deletion prediction a more semantically stable target, because blocks are intentional governance acts while a missing edge can also be a transient network failure.
  • Beyond the paper: the dense, small-world Fedivertex topologies may stress privacy analyses for decentralized learning, since the random-walk privacy budgets reproduced in the paper are tied to communicability and centrality; guarantees derived on sparse synthetic graphs may need rechecking at this density.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 5 minor

Summary. The paper introduces Fedivertex, a graph dataset constructed from seven Fediverse platforms (Mastodon, Misskey, Lemmy, Pleroma, PeerTube, Bookwyrm, Friendica), comprising 13 graph types with 14 weekly snapshots each, totaling 182 graphs. The authors release the dataset on Kaggle, provide a Python package (fedivertex) for convenient NetworkX loading, and publish their crawler code and analysis notebooks. They illustrate the dataset's utility through three applications: reproducing a random-walk differentially private decentralized learning benchmark, a new 'defederation prediction' task for edge and node deletions, and community detection using language labels. The paper claims to be the first publicly released dataset enabling reproducible experiments on Fediverse graphs.

Significance. If its claims are substantiated, Fedivertex is a valuable community resource: it is multi-platform, temporal, openly released with code and notebooks, and it fills a real gap in graph datasets for decentralized machine learning that are not shaped by for-profit recommendation algorithms. The paper is commendably transparent about the failure of its defederation baselines (Table 3, §5.2) and about the robustness of its crawling procedure (Appendix D). The dataset artifact itself is strong and reproducible. However, the 'defederation prediction' task as formally defined is the weakest load-bearing element: it labels every observed edge deletion as defederation, conflating administrative blocks with temporary outages, instance shutdowns, and natural non-recurrence of activity-based edges. Because this task is advertised as a novel contribution in the abstract and in contribution (iii), the manuscript needs substantial revision to either reframe the task or validate the defederation label.

major comments (2)
  1. [§5.2, formal definition of defederation prediction] The formal definition of the defederation task equates 'deleted' edges with all edges in G_t that are absent from G_{t+1} on the common-node subgraph, and the paper calls this phenomenon 'defederation.' In the Fediverse, defederation is an explicit administrative block, typically enacted at the domain level; a weekly snapshot cannot distinguish such a block from a temporary instance outage, a server that stopped responding during the crawl, or an interaction that simply did not recur. This conflation is not a minor modeling choice: it directly affects whether the dataset can support the claimed new task. The paper should either (a) rename the task to 'edge deletion prediction' and discuss its limitations, or (b) obtain ground-truth defederation events (e.g., from admin APIs, blocklists, or moderation logs) to validate that the observed deletions correspond to administrative actions.
  2. [§5.2 and Appendix C] The manuscript's own analysis undermines the defederation-task claim. The authors report that defederation is 'extremely quick' and that weekly granularity yields near-random baseline performance (Table 3: Mastodon AU Del 10 vs. random 6±1.4; Misskey AU Del 1.3–2 vs. random 2±0.8). Appendix C then shows that deletion persistence varies across platforms, with Mastodon and Bookwyrm exhibiting a continuum of durations and Lemmy's activity-based graphs naturally decaying. These observations are consistent with the hypothesis that most week-to-week deletions are not administrative defederations but rather noise from the measurement process. The near-random results therefore do not merely indicate weak baselines; they indicate that the target process is not resolved by the data. Please either acquire higher-frequency snapshots during known defederation periods (as the authors suggest) or clearly state that the current dataset does not support the defederation prediction task as an empirical benchmark.
minor comments (5)
  1. [§3.5] The fourth design principle is introduced with 'Forth,' which should be 'Fourth.'
  2. [§4.2] The sentence 'All the reported graphs exhibit small-world properties to an extend' contains a typo: 'extend' should be 'extent.'
  3. [Table 3] The numeric formatting in Table 3 contains unintended spaces (e.g., '0 .7', '1 .3', '6 ± 1.4'); please fix the table formatting for readability.
  4. [§5.3, Table 4] Community detection results are reported without error bars or statistical significance for the Greedy Modularity and Label Propagation algorithms, and only Louvain is averaged over 100 runs. Given the text's conclusion that 'no method dominates,' please provide variance estimates or state explicitly which entries are single runs.
  5. [Abstract and contribution (iii)] The abstract describes the defederation task as something that 'captures a process of link deletion observed on these networks,' which is accurate, but the term 'defederation' in the task name is likely to be read as a claim about administrative moderation. Please reword the task title and description to avoid overclaiming, even if the dataset remains unchanged.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity; the dataset and benchmarks are self-contained, with only a non-load-bearing self-citation.

full rationale

The derivation chain is self-contained and contains no circular step. The central contribution is the Fedivertex graph collection (Section 3), built by querying instances from the Fediverse Observer list; no parameter is fitted to any target result, and the illustrative experiments do not feed back into the construction. The only self-citation (Section 5.1, reproducing figures of Cyffers et al. [12], co-authored by E. Cyffers) is a benchmark on new graphs, not load-bearing support for the dataset's validity or novelty; even if [12] were wrong, the graph artifact would be unaffected. The defederation task (Section 5.2) defines its target as 'edges either created or deleted' between snapshots; this is a standard supervised label construction, and the name 'defederation' is an interpretive label rather than an input to the prediction. The paper honestly reports that weekly granularity is insufficient ('defederation being extremely quick, and thus the granularity of our current dataset does not seem sufficient to achieve better than random'), which is a measurement-validity limitation rather than circular reasoning. Appendix C's persistence analysis is descriptive post-hoc evidence, not a result derived from the task definition. No uniqueness theorem, ansatz-via-citation, or fitted-input-as-prediction pattern is present.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The central claims rest on dataset construction choices rather than on fitted numerical parameters. The main assumptions are about the representativeness of the Fediverse Observer list and the temporal resolution of the crawler. No free parameters are fitted to a target result, and no new entities are introduced.

assumptions (3)
  • domain assumption Fediverse Observer provides an up-to-date and comprehensive list of instances.
    Section 3.4: crawler relies on this list; if incomplete, graphs miss servers.
  • domain assumption Graph edges capture the intended communication relationships, such as 'interacted at least once' defining a federation edge.
    Section 3.1 and 3.2 define edges; no external ground truth verifies this.
  • domain assumption Weekly snapshots are frequent enough to resolve edge and node deletion dynamics.
    Section 5.2 and Appendix C; the paper itself questions this for fast defederation events.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fedivertex: a Graph Dataset based on Decentralized Social Networks for Trustworthy Machine Learning." pith.science (2026). https://pith.science/paper/FKAFBOFU

@misc{pith2026250520882,
  author       = {Pith},
  title        = {Pith review of: Fedivertex: a Graph Dataset based on Decentralized Social Networks for Trustworthy Machine Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FKAFBOFU}},
  note         = {Machine review of arXiv:2505.20882}
}
read the original abstract

Decentralized machine learning - where each client keeps its own data locally and uses its own computational resources to collaboratively train a model by exchanging peer-to-peer messages - is increasingly popular, as it enables better scalability and control over the data. A major challenge in this setting is that learning dynamics depend on the topology of the communication graph, which motivates the use of real graph datasets for benchmarking decentralized algorithms. Unfortunately, existing graph datasets are largely limited to for-profit social networks crawled at a fixed point in time and often collected at the user scale, where links are heavily influenced by the platform and its recommendation algorithms. The Fediverse, which includes several free and open-source decentralized social media platforms such as Mastodon, Misskey, and Lemmy, offers an interesting real-world alternative. We introduce Fedivertex, a new dataset of 182 graphs, covering seven social networks from the Fediverse, crawled weekly over 14 weeks. We release the dataset along with a Python package to facilitate its use, and illustrate its utility on several tasks, including a new defederation task, which captures a process of link deletion observed on these networks.

Figures

Figures reproduced from arXiv: 2505.20882 by the authors.

Figure 1
Figure 1. Examples of graphs communities based on the official languages in Fedivertex dataset [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Complementary Cumulative Distribution Function (CCDF) of the degree for several [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Numerical experiments reproducing the results of [12] with Fedivertex graphs [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Temporal evolution of Fedivertex graphs evaluated by looking among the top-K predictions how many are indeed new edges, as we report in table 3. Intuitively, deletion could be seen as the opposite of edge creation, so we propose as a baseline, to return the edges with …
Figure 5
Figure 5. Figure 5: Degree-histogram distributions for federation graphs. Version with and without block for [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]
Figure 6
Figure 6. Figure 6: Degree-histogram distributions for the three active-user graphs and Lemmy’s cross- vs. [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: Complementary Cumulative Distribution Function (CCDF) of the degree for several [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Histograms of the language distribution among four different Fediverse software, logarith [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]
Figure 9
Figure 9. Figure 9: Node deletion at different time horizons for federation graphs, and the follow graph (for [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]
Figure 10
Figure 10. Figure 10: Edge deletion at different time horizons for federation graphs, and the follow graph (for [PITH_FULL_IMAGE:figures/full_fig_p017_10.png]
Figure 11
Figure 11. Figure 11: Edge deletion at different time horizons for the three active-user graphs and Lemmy’s [PITH_FULL_IMAGE:figures/full_fig_p018_11.png]
Figure 12
Figure 12. Figure 12: Evolution of the number of nodes over time for the six software of Fedivertex [PITH_FULL_IMAGE:figures/full_fig_p019_12.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

58 extracted references · 57 canonical work pages

  1. [1]

    Decentralised moderation for interoperable social networks: A conversation- based approach for Pleroma and the Fediverse

    Vibhor Agarwal, Aravindh Raman, Nishanth Sastry, Ahmed M Abdelmoniem, Gareth Tyson, and Ignacio Castro. Decentralised moderation for interoperable social networks: A conversation- based approach for Pleroma and the Fediverse. In Conference on Artificial Intelligence (AAAI), volume 18, 2024

    work page 2024

  2. [2]

    Emerging Forms of Sociotechnical Organisation: The Case of the Fediverse

    Jacopo Anderlini and Carlo Milani. Emerging Forms of Sociotechnical Organisation: The Case of the Fediverse. In Emiliana Armano, Marco Briziarelli, and Elisabetta Risi, editors, Digital Platforms and Algorithmic Subjectivities, volume 24. University of Westminster Press, 2022

    work page 2022

  3. [3]

    Network science

    Albert-László Barabási and Márton Pósfai. Network science. Cambridge University Press, Cambridge, 2016

    work page 2016

  4. [4]

    Gephi: an open source software for exploring and manipulating networks

    Mathieu Bastian, Sebastien Heymann, and Mathieu Jacomy. Gephi: an open source software for exploring and manipulating networks. In Proceedings of the international AAAI conference on web and social media, 2009

    work page 2009

  5. [5]

    Personalized and Private Peer-to-Peer Machine Learning

    Aurélien Bellet, Rachid Guerraoui, Mahsa Taziki, and Marc Tommasi. Personalized and Private Peer-to-Peer Machine Learning. In Conference on Uncertainty in Artificial Intelligence (AISTATS), 2018

    work page 2018

  6. [6]

    Toxicity in the Decentralized Web and the Potential for Model Sharing

    Haris Bin Zia, Aravindh Raman, Ignacio Castro, Ishaku Hassan Anaobi, Emiliano De Cristofaro, Nishanth Sastry, and Gareth Tyson. Toxicity in the Decentralized Web and the Potential for Model Sharing. Proc. ACM Meas. Anal. Comput. Syst., 6(2), June 2022

    work page 2022

  7. [7]

    Fast unfolding of communities in large networks

    Vincent D Blondel, Jean-Loup Guillaume, Renaud Lambiotte, and Etienne Lefebvre. Fast unfolding of communities in large networks. Journal of Statistical Mechanics: Theory and Experiment, 2008(10), October 2008

    work page 2008

  8. [8]

    Randomized gossip algorithms

    Stephen Boyd, Arpita Ghosh, Balaji Prabhakar, and Devavrat Shah. Randomized gossip algorithms. IEEE transactions on information theory, 52(6), 2006

    work page 2006

Show all 58 references
  1. [9]

    Bias and debias in recommender system: A survey and future directions

    Jiawei Chen, Hande Dong, Xiang Wang, Fuli Feng, Meng Wang, and Xiangnan He. Bias and debias in recommender system: A survey and future directions. ACM Trans. Inf. Syst., 41(3), February 2023. 10

    2023

  2. [10]

    Aaron Clauset, M. E. J. Newman, and Cristopher Moore. Finding community structure in very large networks. Physical Review E, 70(6), December 2004. Publisher: American Physical Society

    2004

  3. [11]

    Community detection via semi-synchronous label propagation algorithms

    Gennaro Cordasco and Luisa Gargano. Community detection via semi-synchronous label propagation algorithms. In 2010 IEEE International Workshop on: Business Applications of Social Network Analysis (BASNA), December 2010

    2010

  4. [12]

    Differentially private decentralized learning with random walks

    Edwige Cyffers, Aurélien Bellet, and Jalaj Upadhyay. Differentially private decentralized learning with random walks. International Conference on Machine Learning (ICML), 2024

    2024

  5. [13]

    Muffliato: Peer-to- peer privacy amplification for decentralized optimization and averaging

    Edwige Cyffers, Mathieu Even, Aurélien Bellet, and Laurent Massoulié. Muffliato: Peer-to- peer privacy amplification for decentralized optimization and averaging. Advances in Neural Information Processing Systems (NeurIPS), 35, 2022

    2022

  6. [14]

    Fedivertex, 2025

    Marc Damie and Edwige Cyffers. Fedivertex, 2025. URL: https://www.kaggle.com/ds/6877842

    2025

  7. [15]

    Dekker, Zekeriya Erkin, and Mauro Conti

    Florine W. Dekker, Zekeriya Erkin, and Mauro Conti. Topology-based reconstruction prevention for decentralised learning, 2023

    2023

  8. [16]

    Dunlavy, Tamara G

    Daniel M. Dunlavy, Tamara G. Kolda, and Evrim Acar. Temporal link prediction using matrix and tensor factorizations. ACM Trans. Knowl. Discov. Data, 5(2), February 2011

    2011

  9. [17]

    Stochastic Gradient Descent under Markovian Sampling Schemes, February

    Mathieu Even. Stochastic Gradient Descent under Markovian Sampling Schemes, February

  10. [18]

    Lee Giles, Kurt D

    C. Lee Giles, Kurt D. Bollacker, and Steve Lawrence. Citeseer: an automatic citation indexing system. In Proceedings of the Third ACM Conference on Digital Libraries, DL ’98, New York, NY , USA, 1998. Association for Computing Machinery

    1998

  11. [19]

    Decentralized Federated Learning: A Survey on Security and Privacy

    Ehsan Hallaji, Roozbeh Razavi-Far, Mehrdad Saif, Boyu Wang, and Qiang Yang. Decentralized Federated Learning: A Survey on Security and Privacy. IEEE Transactions on Big Data, 10(2), April 2024

    2024

  12. [20]

    Mohammad Al Hasan and Mohammed J. Zaki. A Survey of Link Prediction in Social Networks. In Charu C. Aggarwal, editor, Social Network Data Analytics. Springer US, Boston, MA, 2011

    2011

  13. [21]

    A principled framework for the design and analysis of token algorithms, May 2022

    Hadrien Hendrikx. A principled framework for the design and analysis of token algorithms, May 2022. Number: arXiv:2205.15015 arXiv:2205.15015 [cs, math]

    2022 arXiv

  14. [22]

    Dual-free stochastic decentralized optimization with variance reduction

    Hadrien Hendrikx, Francis Bach, and Laurent Massoulié. Dual-free stochastic decentralized optimization with variance reduction. In Conference on Neural Information Processing Systems (NeurIPS), volume 33. Curran Associates, Inc., 2020

    2020

  15. [23]

    Comparing user activity on X and Mastodon

    Shiori Hironaka, Mitsuo Yoshida, and Kazuyuki Shudo. Comparing user activity on X and Mastodon. In 2024 IEEE International Conference on Big Data (BigData). IEEE, December 2024

    2024

  16. [24]

    The Non-IID Data Quagmire of Decentralized Machine Learning

    Kevin Hsieh, Amar Phanishayee, Onur Mutlu, and Phillip Gibbons. The Non-IID Data Quagmire of Decentralized Machine Learning. In International Conference on Machine Learing (ICML). PMLR, November 2020. ISSN: 2640-3498

    2020

  17. [25]

    Russell Bernard, and Huan Liu

    Ujun Jeong, Paras Sheth, Anique Tahir, Faisal Hammad Alatawi, H. Russell Bernard, and Huan Liu. Exploring platform migration patterns between Twitter and Mastodon: A user behavior study, 2023

    2023

  18. [26]

    A randomized incremental subgradient method for distributed optimization in networked systems

    Björn Johansson, Maben Rabi, and Mikael Johansson. A randomized incremental subgradient method for distributed optimization in networked systems. SIAM Journal on Optimization , 20(3), 2010

    2010

  19. [27]

    Brendan McMahan, Brendan Avent, Aurélien Bellet, Mehdi Bennis, Ar- jun Nitin Bhagoji, Kallista Bonawitz, Zachary Charles, Graham Cormode, Rachel Cummings, Rafael G

    Peter Kairouz, H. Brendan McMahan, Brendan Avent, Aurélien Bellet, Mehdi Bennis, Ar- jun Nitin Bhagoji, Kallista Bonawitz, Zachary Charles, Graham Cormode, Rachel Cummings, Rafael G. L. D’Oliveira, Hubert Eichner, Salim El Rouayheb, David Evans, Josh Gardner, 11 Zachary Garret...

    2021

  20. [28]

    Impact of Network Topology on the Convergence of Decentralized Federated Learning Systems

    Hanna Kavalionak, Emanuele Carlini, Patrizio Dazzi, Luca Ferrucci, Matteo Mordacchini, and Massimo Coppola. Impact of Network Topology on the Convergence of Decentralized Federated Learning Systems. In 2021 IEEE Symposium on Computers and Communications (ISCC), September 2021....

    2021

  21. [29]

    A unified theory of decentralized SGD with changing topology and local updates

    Anastasia Koloskova, Nicolas Loizou, Sadra Boreiri, Martin Jaggi, and Sebastian Stich. A unified theory of decentralized SGD with changing topology and local updates. In International Conference on Machine Learing (ICML), volume 119 of Proceedings of Machine Learning Research....

    2020

  22. [30]

    New Paradigms in Trust and Safety: Navigating Defederation on Decentralized Social Media Platforms

    Samantha Lai, Yoel Roth, Kate Klonick, Mallory Knodel, Evan Prodromou, and Aaron Roder- icks. New Paradigms in Trust and Safety: Navigating Defederation on Decentralized Social Media Platforms. Technical report, Carnegie Endowment for International Peace, April 2025

    2025

  23. [31]

    Refined convergence and topology learning for decentralized sgd with heterogeneous data

    Batiste Le Bars, Aurélien Bellet, Marc Tommasi, Erick Lavoie, and Anne-Marie Kermarrec. Refined convergence and topology learning for decentralized sgd with heterogeneous data. In Conference on Uncertainty in Artificial Intelligence (AISTATS), volume 206 of Proceedings of Mach...

    2023

  24. [32]

    Activitypub w3c recommendation, 2018

    Christine Lemmer-Webber, Jessica Tallon, Erin Shepherd, Amy Guy, and Evan Prodromou. Activitypub w3c recommendation, 2018

    2018

  25. [33]

    Predicting positive and negative links in online social networks

    Jure Leskovec, Daniel Huttenlocher, and Jon Kleinberg. Predicting positive and negative links in online social networks. In Proceedings of the 19th international conference on World wide web, WWW ’10, New York, NY , USA, April 2010. Association for Computing Machinery

    2010

  26. [34]

    SNAP Datasets: Stanford large network dataset collection, 2014

    Jure Leskovec and Andrej Krevl. SNAP Datasets: Stanford large network dataset collection, 2014

    2014

  27. [35]

    Learning to Discover Social Circles in Ego Networks

    Jure Leskovec and Julian Mcauley. Learning to Discover Social Circles in Ego Networks. In Conference on Neural Information Processing Systems (NeurIPS), volume 25. Curran Asso- ciates, Inc., 2012

    2012

  28. [36]

    Federated learning: Chal- lenges, methods, and future directions

    Tian Li, Anit Kumar Sahu, Ameet Talwalkar, and Virginia Smith. Federated learning: Chal- lenges, methods, and future directions. IEEE Signal Processing Magazine, 37(3), May 2020

    2020

  29. [37]

    Can decentralized algorithms outperform centralized algorithms? a case study for decentralized parallel stochastic gradient descent

    Xiangru Lian, Ce Zhang, Huan Zhang, Cho-Jui Hsieh, Wei Zhang, and Ji Liu. Can decentralized algorithms outperform centralized algorithms? a case study for decentralized parallel stochastic gradient descent. In Conference on Neural Information Processing Systems (NeurIPS), NIPS...

    2017

  30. [38]

    The link-prediction problem for social networks

    David Liben-Nowell and Jon Kleinberg. The link-prediction problem for social networks. Journal of the American Society for Information Science and Technology, 58(7), March 2007

    2007

  31. [39]

    Dynamics on higher-order networks: a review

    Soumen Majhi, Matjaž Perc, and Dibakar Ghosh. Dynamics on higher-order networks: a review. Journal of The Royal Society Interface, 19(188), March 2022

    2022

  32. [40]

    Sayed, and Wotao Yin

    Xianghui Mao, Kun Yuan, Yubin Hu, Yuantao Gu, Ali H. Sayed, and Wotao Yin. Walkman: A Communication-Efficient Random-Walk Algorithm for Decentralized Optimization. IEEE Transactions on Signal Processing, 68, 2020

    2020

  33. [41]

    Measuring and moderating opinion polarization in social networks

    Antonis Matakos, Evimaria Terzi, and Panayiotis Tsaparas. Measuring and moderating opinion polarization in social networks. Data Mining and Knowledge Discovery, 31(5), July 2017. 12

    2017

  34. [42]

    Automating the construction of internet portals with machine learning

    Andrew Kachites McCallum, Kamal Nigam, Jason Rennie, and Kristie Seymore. Automating the construction of internet portals with machine learning. Information Retrieval, 3(2), 2000

    2000

  35. [43]

    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. In Conference on Uncertainty in Artificial Intelligence (AISTATS) , volume 54 of Proceedings of Machine Learning Res...

    2017

  36. [44]

    Dsa: Decentralized double stochastic averaging gradient algorithm

    Aryan Mokhtari and Alejandro Ribeiro. Dsa: Decentralized double stochastic averaging gradient algorithm. Journal of Machine Learning Research, 17(61), 2016

    2016

  37. [45]

    Privacy attacks in decentralized learning

    Abdellah El Mrini, Edwige Cyffers, and Aurélien Bellet. Privacy attacks in decentralized learning. International Conference on Machine Learning (ICML), 2024

    2024

  38. [46]

    Decentralized gradient methods: does topology matter? In Conference on Uncertainty in Artificial Intelligence (AISTATS), 2020

    Giovanni Neglia, Chuan Xu, Don Towsley, and Gianmarco Calbi. Decentralized gradient methods: does topology matter? In Conference on Uncertainty in Artificial Intelligence (AISTATS), 2020

    2020

  39. [47]

    Bronstein

    Emanuele Rossi, Benjamin Paul Chamberlain, Fabrizio Frasca, Davide Eynard, Federico Monti, and Michael M. Bronstein. Temporal graph networks for deep learning on dynamic graphs, 2020

    2020

  40. [48]

    Multi-scale attributed node embedding, 2019

    Benedek Rozemberczki, Carl Allen, and Rik Sarkar. Multi-scale attributed node embedding, 2019

    2019

  41. [49]

    Can this platform survive? governance challenges for the Fediverse

    Thomas Struett, Aram Sinnreich, Patricia Aufderheide, and Robert Gehl. Can this platform survive? governance challenges for the Fediverse. SSRN Electronic Journal, 2023

    2023

  42. [50]

    The effect of recommendations on network structure

    Jessica Su, Aneesh Sharma, and Sharad Goel. The effect of recommendations on network structure. In Proceedings of the 25th International Conference on World Wide Web, WWW ’16, Republic and Canton of Geneva, CHE, 2016. International World Wide Web Conferences Steering Committee

    2016

  43. [51]

    D2: Decentralized Training over Decentralized Data

    Hanlin Tang, Xiangru Lian, Ming Yan, Ce Zhang, and Ji Liu. D2: Decentralized Training over Decentralized Data. In International Conference on Machine Learing (ICML), 2018

    2018

  44. [52]

    Graph Mining Applications to Social Network Analysis

    Lei Tang and Huan Liu. Graph Mining Applications to Social Network Analysis. In Charu C. Aggarwal and Haixun Wang, editors, Managing and Mining Graph Data. Springer US, Boston, MA, 2010

    2010

  45. [53]

    A comparative study of social network models: Network evolution models and nodal attribute models

    Riitta Toivonen, Lauri Kovanen, Mikko Kivelä, Jukka-Pekka Onnela, Jari Saramäki, and Kimmo Kaski. A comparative study of social network models: Network evolution models and nodal attribute models. Social Networks, 31(4), October 2009

    2009

  46. [54]

    RelaySum for Decentralized Deep Learning on Heterogeneous Data

    Thijs V ogels, Lie He, Anastasiia Koloskova, Sai Praneeth Karimireddy, Tao Lin, Sebastian U Stich, and Martin Jaggi. RelaySum for Decentralized Deep Learning on Heterogeneous Data. In Advances in Neural Information Processing Systems, volume 34. Curran Associates, Inc., 2021

    2021

  47. [55]

    A Mastodon Corpus to Evaluate Federated Microblog Search

    Matti Wiegmann, Jan Heinrich Reimer, Maximilian Ernst, Martin Potthast, Matthias Hagen, and Benno Stein. A Mastodon Corpus to Evaluate Federated Microblog Search. In WOWS@ ECIR, pages 37–49, 2024

    2024

  48. [56]

    An information flow model for conflict and fission in small groups

    Wayne Zachary. An information flow model for conflict and fission in small groups. Journal of anthropological research, 33, 11 1976

    1976

  49. [57]

    Mastodon

    Matteo Zignani, Sabrina Gaito, and Gian Paolo Rossi. Follow the “Mastodon”: Structure and Evolution of a Decentralized Online Social Network. Proceedings of the International AAAI Conference on Web and Social Media, 12(1), June 2018. Number: 1. 13 A Further analysis of Fediver...

    2018

  50. [2023]

    arXiv:2302.14428 [cs, math]

    arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.