Generated Text Detection using Representation Learning with Token Prediction Model 


Vol. 51,  No. 2, pp. 241-247, Feb.  2026
10.7840/kics.2026.51.2.241


PDF Full-Text
  Abstract

Large language models (LLMs) have become mainstream in the form of conversational chatbots, such as ChatGPT, and various social problems have begun to arise. Representative issues include AI plagiarism where individuals claim AI-generated content as their own and fake news creation and dissemination. To prevent the misuse of AI-generated text, several methods have been proposed to distinguish AI-generated content from human-written text. Popular approaches rely on zero-shot detection, which compares the negative probability curvature derived from token prediction results using various LLMs. More recently, learning-based methods have been proposed to replicate the output distribution of the target model for easier adaptation to various generators. However, these existing methods primarily depend on token-level probability estimation, making them vulnerable to variations in sentence expression, such as paraphrasing. In this paper, we propose a novel approach that applies representation learning to the prediction model to overcome these limitations. The proposed method aims to enhance robustness against paraphrasing and expression diversity, and we demonstrate its potential effectiveness through empirical evaluation.

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Related Articles
  Cite this article

[IEEE Style]

Y. Choo and Y. Park, "Generated Text Detection using Representation Learning with Token Prediction Model," The Journal of Korean Institute of Communications and Information Sciences, vol. 51, no. 2, pp. 241-247, 2026. DOI: 10.7840/kics.2026.51.2.241.

[ACM Style]

Yeon-Seung Choo and Yong-Suk Park. 2026. Generated Text Detection using Representation Learning with Token Prediction Model. The Journal of Korean Institute of Communications and Information Sciences, 51, 2, (2026), 241-247. DOI: 10.7840/kics.2026.51.2.241.

[KICS Style]

Yeon-Seung Choo and Yong-Suk Park, "Generated Text Detection using Representation Learning with Token Prediction Model," The Journal of Korean Institute of Communications and Information Sciences, vol. 51, no. 2, pp. 241-247, 2. 2026. (https://doi.org/10.7840/kics.2026.51.2.241)
Vol. 51, No. 2 Index

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880