We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Cryptography and Security

Title: A Word is Worth A Thousand Dollars: Adversarial Attack on Tweets Fools Stock Predictions

Abstract: More and more investors and machine learning models rely on social media (e.g., Twitter and Reddit) to gather real-time information and sentiment to predict stock price movements. Although text-based models are known to be vulnerable to adversarial attacks, whether stock prediction models have similar vulnerability is underexplored. In this paper, we experiment with a variety of adversarial attack configurations to fool three stock prediction victim models. We address the task of adversarial generation by solving combinatorial optimization problems with semantics and budget constraints. Our results show that the proposed attack method can achieve consistent success rates and cause significant monetary loss in trading simulation by simply concatenating a perturbed but semantically similar tweet.
Comments: NAACL short paper, github: this https URL
Subjects: Cryptography and Security (cs.CR); Machine Learning (cs.LG); Statistical Finance (q-fin.ST)
Cite as: arXiv:2205.01094 [cs.CR]
  (or arXiv:2205.01094v3 [cs.CR] for this version)

Submission history

From: Yong Xie [view email]
[v1] Sun, 1 May 2022 05:12:22 GMT (1662kb,D)
[v2] Wed, 11 May 2022 02:19:10 GMT (1630kb,D)
[v3] Tue, 12 Jul 2022 19:09:00 GMT (3320kb,D)

Link back to: arXiv, form interface, contact.