A framework for differentially-private knowledge graph embeddings

Han, Xiaolin; Dell’Aglio, Daniele; Grubenmann, Tobias; Cheng, Reynold; Bernstein, Abraham

doi:10.1016/j.websem.2021.100696

Cited by 12 publications

(8 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[ 92 ] indicated that the green investment market facilitates the mobilization of debt money for low CO 2 emission projects and investigated the impact of green investment on the short- and long-term performance of IPOs (IPOS) [ 93 ]. also illustrated that investor interest is higher and company performance is higher for those with green investments [ 7 , 76 , 94 , 95 ]. have analyzed the primary energy sources and renewable energy usage and discussed that the CO 2 emissions in China are the greatest in the world, according to the World Population Review.…”

Section: Literature Reviewmentioning

confidence: 99%

“…CO 2 emissions are broken down into traded and untraded components and compared to underproductive, consumption-based emissions. Through the use of structural decomposition analysis, they uncovered what exactly causes the intensity of emissions to shift [ 75 , 76 ]. Cultural economic indicators, which pertain to how different societies recorded their resource production and consumption, were used to illustrate the economy.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Catch-up growth with alpha and beta decoupling and their relationships between CO2 emissions by GDP, population, energy production, and consumption

Khan

2024

Heliyon

View full text Add to dashboard Cite

Section: Literature Reviewmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Catch-up growth with alpha and beta decoupling and their relationships between CO2 emissions by GDP, population, energy production, and consumption

Khan

2024

Heliyon

View full text Add to dashboard Cite

“…Han et al [17] propose a DPSGD-based differentially private KGE algorithm for a single knowledge graph in the centralized learning setting [3]. Although they also attempt to only perturb the active gradient elements binding to the sampled entities, they fail to conduct the selection in a differentially private manner.…”

Section: H3 Inaccurate Dp Analysis Of Existing Workmentioning

confidence: 99%

“…DP has been investigated for KGE privacy protection. Han et al [17] propose a DPSGD-based DP KGE algorithm for a single KG in the centralized learning. Peng et al [30] propose a PATE-GAN-based DP KGE algorithm for multiple KGs in the decentralized learning with the off-the-shelf PATE-GAN technique [21].…”

Section: A Score Functionsmentioning

confidence: 99%

“…Further, inspired by the varying attack strengths across the FKGE training procedure, we design an adaptive privacy budget allocation strategy to adjust the defense magnitude accordingly. Finally, although differential privacy has been investigated for KGE privacy protection in both centralized [17] and federated settings [30], existing attempts are not accurate enough with privacy analysis, which renders their defense mechanism not rigorous in differential privacy guarantee. Please refer to Appendix H.3 for further discussion.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Quantifying and Defending against Privacy Threats on Federated Knowledge Graph Embedding

Liang

Wu³

et al. 2023

Proceedings of the ACM Web Conference 2023

View full text Add to dashboard Cite

Knowledge Graph Embedding (KGE) is a fundamental technique that extracts expressive representation from knowledge graph (KG) to facilitate diverse downstream tasks. The emerging federated KGE (FKGE) collaboratively trains from distributed KGs held among clients while avoiding exchanging clients' sensitive raw KGs, which can still suffer from privacy threats as evidenced in other federated model trainings (e.g., neural networks). However, quantifying and defending against such privacy threats remain unexplored for FKGE which possesses unique properties not shared by previously studied models. In this paper, we conduct the first holistic study of the privacy threat on FKGE from both attack and defense perspectives. For the attack, we quantify the privacy threat by proposing three new inference attacks, which reveal substantial privacy risk by successfully inferring the existence of the KG triple from victim clients. For the defense, we propose DP-Flames, a novel differentially private FKGE with private selection, which offers a better privacy-utility tradeoff by exploiting the entity-binding sparse gradient property of FKGE and comes with a tight privacy accountant by incorporating the state-of-the-art private selection technique. We further propose an adaptive privacy budget allocation policy to dynamically adjust defense magnitude across the training procedure. Comprehensive evaluations demonstrate that the proposed defense can successfully mitigate the privacy threat by effectively reducing the success rate of inference attacks from 83.1% to 59.4% on average with only a modest utility decrease.

show abstract

TRoute: Dynamic Time-Dependent Route Recommendation on Road Networks

Han,

Hu,

et al. 2024

Lecture Notes in Computer Science

View full text Add to dashboard Cite

A framework for differentially-private knowledge graph embeddings

Cited by 12 publications

References 36 publications

Catch-up growth with alpha and beta decoupling and their relationships between CO2 emissions by GDP, population, energy production, and consumption

Catch-up growth with alpha and beta decoupling and their relationships between CO2 emissions by GDP, population, energy production, and consumption

Quantifying and Defending against Privacy Threats on Federated Knowledge Graph Embedding

TRoute: Dynamic Time-Dependent Route Recommendation on Road Networks

Contact Info

Product

Resources

About