Adversarial Attacks on Knowledge Graph Embeddings via Instance Attribution Methods

Bhardwaj, Peru; Kelleher, John D.; Costabello, Luca; O’Sullivan, Declan

doi:10.18653/v1/2021.emnlp-main.648

Cited by 12 publications

(3 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To address this, we adopt an estimate approach inspired by the Influence Function. 50,51 We first upweight z with a small weight ε and define the new optimal embeddings as . We then calculate the impact of adding z m on as follows: where is the Hessian matrix, computed as .…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Poisoning scientific knowledge using large language models

Yang,

Xu,

Mirzoyan

et al. 2023

Preprint

View full text Add to dashboard Cite

Biomedical knowledge graphs constructed from scientific literature have been widely used to validate biological discoveries and generate new hypotheses. Recently, large language models (LLMs) have demonstrated a strong ability to generate human-like text data. While most of these text data have been useful, LLM might also be used to generate malicious content. Here, we investigate whether it is possible that a malicious actor can use LLM to generate a malicious paper that poisons scientific knowledge graphs and further affects downstream biological applications. As a proof-of-concept, we develop Scorpius, a conditional text generation model that generates a malicious paper abstract conditioned on a promoting drug and a target disease. The goal is to fool the knowledge graph constructed from a mixture of this malicious abstract and millions of real papers so that knowledge graph consumers will misidentify this promoting drug as relevant to the target disease. We evaluated Scorpius on a knowledge graph constructed from 3,818,528 papers and found that Scorpius can increase the relevance of 71.3% drug disease pairs from the top 1000 to the top 10 by only adding one malicious abstract. Moreover, the generation of Scorpius achieves better perplexity than ChatGPT, suggesting that such malicious abstracts cannot be efficiently detected by humans. Collectively, Scorpius demonstrates the possibility of poisoning scientific knowledge graphs and manipulating downstream applications using LLMs, indicating the importance of accountable and trustworthy scientific knowledge discovery in the era of LLM.

show abstract

Section: Methodsmentioning

confidence: 99%

“…To address this, we adopt an estimate approach inspired by the Influence Function. 50,51 We first upweight with a small weight and define the 𝑧 .…”

Section: Selecting Poisoning Targetmentioning

confidence: 99%

Poisoning scientific knowledge using large language models

Yang,

Xu,

Mirzoyan

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Description: An adversarial attack against knowledge graph embedding aims at identifying the training instances that are most influential to the model's predictions on test instances. Existing works in this area are limited (Bhardwaj et al 2021;Betz, Meilicke, and Stuckenschmidt 2022), and even more limited is the design of a defense mechanism to alleviate the effect of adversarial attacks against knowledge graph embedding methods.…”

Section: Cross Domain Clusteringmentioning

confidence: 99%

Learning Better Representations Using Auxiliary Knowledge

Rezayi¹

2023

AAAI

View full text Add to dashboard Cite

Representation Learning is the core of Machine Learning and Artificial Intelligence as it summarizes input data points into low dimensional vectors. This low dimensional vectors should be accurate portrayals of the input data, thus it is crucial to find the most effective and robust representation possible for given input as the performance of the ML task is dependent on the resulting representations. In this summary, we discuss an approach to augment representation learning which relies on external knowledge. We briefly describe the shortcoming of the existing techniques and describe how an auxiliary knowledge source could result in obtaining improved representations.

show abstract