Přístupnostní navigace
E-přihláška
Vyhledávání Vyhledat Zavřít
Detail publikace
ČEGIŇ, J.
Originální název
A Game for Crowdsourcing Adversarial Examples for False Information Detection
Typ
článek ve sborníku ve WoS nebo Scopus
Jazyk
angličtina
Originální abstrakt
False information detection models are susceptible to adversarial attacks. Such susceptibility is a critical weakness of detection models. Automated creation of adversarial samples can ultimately help to augment training sets and create more robust detection models. However, automatically generated adversarial samples often do not preserve the meaning contained in the original text, leading to information loss. There is a need for adversarial sample generators that can preserve the original meaning. To explore the properties such generators should have and to inform their future design, we conducted a study to collect adversarial samples from human agents using a Game with a purpose (GWAP). Players goal is to modify a given tweet until a detection model is tricked thus creating an adversarial sample. We qualitatively analysed the collected adversarial samples and identified desired properties/strategies that an adversarial meaning-preserving generator should exhibit. These strategies are validated on detection models based on a transformer and LSTM models to confirm their applicability on different models. Based on these findings, we propose a novel generator approach that will exhibit the desired properties in order to generate high-quality information-preserving adversarial samples.
Klíčová slova
adversarial data generation, false information detection, game with a purpose, human interaction task, machine learning
Autoři
Vydáno
1. 12. 2022
Nakladatel
CEUR-WS.org
Místo
Vídeň
ISSN
1613-0073
Periodikum
CEUR Workshop Proceedings
Ročník
2022
Číslo
Stát
Spolková republika Německo
Strany od
13
Strany do
25
Strany počet
17
URL
https://ceur-ws.org/Vol-3275/paper2.pdf
BibTex
@inproceedings{BUT182948, author="Ján {Čegiň}", title="A Game for Crowdsourcing Adversarial Examples for False Information Detection", booktitle="CEUR Workshop Proceedings", year="2022", journal="CEUR Workshop Proceedings", volume="2022", number="2022", pages="13--25", publisher="CEUR-WS.org", address="Vídeň", issn="1613-0073", url="https://ceur-ws.org/Vol-3275/paper2.pdf" }
Dokumenty
_AIofAI__A_Game_for_Crowdsourcing_Adversarial_Examples_for_False_Information_Detection_subm.pdf