SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.13862 · NeurIPS · 2025

Competitive Advantage Attacks to Decentralized Federated Learning

Neil Gong, Minghong Fang, Yuqi Jia

arXiv · PDF · Open in the Atlas

Code that ran

1 of 1 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 1 failure of our instrument; 0 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2310.13862.json · badge: /api/ran/2310.13862.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
jyqhahah/SelfishAttack community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
selfish_fedavg Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jyqhahah/SelfishAttack/aggregation.py
pointer only (licence: NONE) · get_code("91f218d0865d736b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jyqhahah/SelfishAttack when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Also present in other papers' code

Of 1 samples lifted from this paper's linked repositories, 1 carry a function identity and 0 do not. Of the 1 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 1 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Decentralized federated learning (DFL) enables clients (e.g., hospitals and banks) to jointly train machine learning models without a central orchestration server. In each global training round, each client trains a local model on its own training data and then they exchange local models for aggregation. In this work, we propose SelfishAttack, a new family of attacks to DFL. In SelfishAttack, a set of selfish clients aim to achieve competitive advantages over the remaining nonselfish ones, i.e., the final learnt local models of the selfish clients are more accurate than those of the non-selfish ones. Towards this goal, the selfish clients send carefully crafted local models to each remaining non-selfish one in each global training round. We formulate finding such local models as an optimization problem and propose methods to solve it when DFL uses different aggregation rules. Theoretically, we show that our methods find the optimal solutions to the optimization problem. Empirically, we show that SelfishAttack successfully increases the accuracy gap (i.e., competitive advantage) between the final learnt local models of selfish clients and those of non-selfish ones. Moreover, SelfishAttack achieves larger accuracy gaps than poisoning attacks when extended to increase competitive advantages. Our code and data are available at: https://github.com/jyqhahah/SelfishAttack. To defend against adversarial behaviors, several Byzantine-robust aggregation rules have been proposed, including Median [36], Trimmed-mean [36], and Krum [3], which aim to limit the influence of outlier models. More advanced defenses [5,27,11,25,14,15,7,30,19,37,12] like FLTrust [5], FLDetector [37], RFA [27], and FLAME [25] use additional mechanisms such as trust scoring or clustering. We summarize these defenses in Appendix A.3. Attacker's goal: We consider a colluding subset of clients, termed selfish clients, as the attacker. Their goal is to compromise the DFL training to achieve two objectives: 1). The utility goal: selfish clients aim to obtain more accurate local models than those learned by training solely among themselves. Formally, let M 1 be the local model obtained when selfish clients train solely among themselves, and M 2 the model learned when they participate in full DFL. The utility goal requires that M 2 is more accurate than M 1 . 2). The competitive-advantage goal: the selfish clients aim to obtain more accurate local models than those of the non-selfish clients, gaining practical benefits such as improved service quality or increased revenue in competitive domains. Attacker's capabilities: A selfish client sends its true pre-aggregation local model to other selfish clients but can arbitrarily manipulate the models shared with non-selfish clients. In contrast, nonselfish clients follow standard DFL behavior and share their pre-aggregation local models with all peers. We assume the standard resilience condition N ≥ 3m + 1 [21], where m is the number of selfish clients and N is the total number of clients. Fig. 1 illustrates the attack mechanism. Attacker's knowledge: As legitimate participants, selfish clients know the aggregation rule and receive the pre-aggregation local models from all non-selfish clients. They also share their own pre-aggregation local models among themselves. Comparison with poisoning attacks: Unlike traditional poisoning attacks [3,6,10], which typically involve external adversaries injecting faulty updates to indiscriminately degrade model performance, SelfishAttack is an internal, coordinated attack. Poisoning attacks often reduce the performance of all involved clients, including compromised ones, yielding no competitive advantage. In contrast, SelfishAttack carefully manipulates only selected updates to degrade non-selfish clients' models while maintaining or improving the accuracy of selfish clients' own models. As shown in our experiments, compared to poisoning attacks, SelfishAttack effectively degrades non-selfish clients' performance under robust aggregation, while also giving selfish clients a competitive advantage.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2310.13862")
get_code_for_paper("2310.13862")
have("2310.13862")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2310.13862.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2310.13862.svg)](https://syntology.ai/paper/2310.13862)

Connect an agent — have() is free.