Adversarial Attacks on Knowledge Graph Embeddings via Instance Attribution Methods (2021.emnlp-main)
Copied to clipboard
| Challenge: | Knowledge Graph Embeddings (KGE) are widely used for relational learning on large scale Knowledge . however, little is known about the security vulnerabilities that might disrupt their intended behaviour. |
| Approach: | They propose to use model-agnostic instance attribution methods to select adversarial deletions and a heuristic method to replace one of the two entities in each influential triple to generate adversarials. |
| Outcome: | The proposed methods outperform the state-of-the-art data poisoning attacks on KGE models and improve the MRR degradation by up to 62% over the baselines. |
Similar Papers
Poisoning Knowledge Graph Embeddings via Relation Inference Patterns (2021.acl-long)
Copied to clipboard
| Challenge: | Knowledge graph embeddings (KGE) models are increasingly deployed in domains with high stake decision making where it is critical to identify the potential security vulnerabilities that might cause failure. |
| Approach: | They propose to exploit the inductive abilities of knowledge graph embedding models by crafting adversarial additions that can improve model’s confidence on decoy facts. |
| Outcome: | The proposed attacks outperform state-of-the-art baselines on four KGE models for two publicly available datasets and generalize across all model-dataset combinations. |
KBGAN: Adversarial Learning for Knowledge Graph Embeddings (N18-1)
Copied to clipboard
| Challenge: | Existing knowledge graph embedding techniques lack the capability to access similarities between entities and relations. |
| Approach: | They propose an adversarial learning framework to improve knowledge graph embedding models . they use one knowledge graph embedded model as a negative sample generator . |
| Outcome: | The proposed framework improves the performance of knowledge graph embedding models on a link prediction task. |
KGE-CL: Contrastive Learning of Tensor Decomposition Based Knowledge Graph Embeddings (2022.coling-1)
Copied to clipboard
| Challenge: | Existing knowledge graph embedding methods ignore semantic similarity between related entities and entity-relation couples in different triples . |
| Approach: | They propose a contrastive learning framework for tensor decomposition based (TDB) KGE that can shorten the semantic distance of related entities and entity-relation couples in different triples and thus improve the performance of KGE. |
| Outcome: | The proposed method achieves 51.2% MRR, 46.8% Hits@1 on three standard KGE datasets, 37.8% MRR and 28.6% Hits @1 on FB15k-237 datasets and 59.1% MRR . |
Adversarial Attack against Cross-lingual Knowledge Graph Alignment (2021.emnlp-main)
Copied to clipboard
Zeru Zhang, Zijie Zhang, Yang Zhou, Lingfei Wu, Sixing Wu, Xiaoying Han, Dejing Dou, Tianshi Che, Da Yan
| Challenge: | Existing studies on cross-lingual entity alignment under adversarial attacks have not been conducted. |
| Approach: | They propose to use adversarial attack techniques to perturb cross-lingual entity alignment under adversarials. |
| Outcome: | The proposed model hides the attacked entities in dense regions in two KGs, and reduces the gradient vanishing issues in the process of adversarial attacks for further improving the attack effectiveness. |
Investigating Robustness and Interpretability of Link Prediction via Adversarial Modifications (N19-1)
Copied to clipboard
| Challenge: | Existing approaches focus on improving accuracy and overlook other aspects such as robustness and interpretability. |
| Approach: | They propose adversarial modifications for link prediction models that identify influential facts and evaluate their sensitivity to addition of fake facts. |
| Outcome: | The proposed model evaluates the robustness of the model to the addition of fake facts and the interpretability of the models. |
Debiasing knowledge graph embeddings (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to train knowledge graph embeddings to be neutral to sensitive attributes such as gender have been shown to increase training time by a factor of eight or more. |
| Approach: | They propose a method where all embeddings are trained to be neutral to sensitive attributes such as gender by default using an adversarial loss. |
| Outcome: | The proposed method reduces training time by eightfold and improves accuracy. |
KGE Calibrator: An Efficient Probability Calibration Method of Knowledge Graph Embedding Models for Trustworthy Link Prediction (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for probability calibration of knowledge graph embedding models are ill-suited for KGEs. |
| Approach: | They propose a method to calibrate knowledge graph embedding models for ranking-based link prediction using a Jump Selection Strategy and Multi-Binning Scaling to enhance reliability. |
| Outcome: | Experiments show that the KGEC outperforms existing calibration methods in terms of effectiveness and efficiency. |
Concealed Data Poisoning Attacks on NLP Models (2021.naacl-main)
Copied to clipboard
| Challenge: | In contrast, adversarial attacks can cause model errors by modifying inputs, such as the universal triggers attack. |
| Approach: | They propose a data poisoning attack that allows an adversary to control model predictions whenever a desired trigger phrase is present in the input. |
| Outcome: | The proposed attack can cause model errors by modifying inputs, but it can also cause extra human annotation. |
Predictive Multiplicity of Knowledge Graph Embeddings in Link Prediction (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Knowledge graph embeddings (KGE) models are often used to predict missing links for knowledge graphs (KGs) however, multiple KG embedds can give conflicting predictions for unseen queries. |
| Approach: | They define predictive multiplicity in link prediction and introduce evaluation metrics to measure it using commonly used benchmark datasets. |
| Outcome: | The proposed methods significantly mitigat conflicts by 66% to 78% in link prediction. |
Evaluating the Calibration of Knowledge Graph Embeddings for Trustworthy Link Prediction (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing calibration techniques are less effective under the standard closed-world assumption (CWA) and the more realistic open-world hypothesis (OWA) Existing methods are not effective under OWA and provide explanations for this discrepancy. |
| Approach: | They conduct an evaluation under the standard closed-world assumption (CWA) and introduce the more realistic but challenging open-world assume (OWA) . they find existing calibration techniques are much less effective under the OWA than the CWA . |
| Outcome: | The proposed calibration techniques are much less effective under the open-world assumption (OWA) and explain the discrepancy. |