Detecting Fake Content with Relative Entropy Scoring
By Thomas Lavergne, Tanguy Urvoy, François Yvon. 2008
111 citations recorded by the source corpus.
Subfields: Natural Language Processing, Safety, Robustness & Interpretability
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Natural Language Processing
- Mohit Iyyer
- Sebastian Gehrmann
- John Wieting
- Kalpesh Krishna
- Graham Neubig
- Yonatan Belinkov
- Alexander M. Rush
- Mohit Bansal
In Safety, Robustness & Interpretability
- Sebastian Gehrmann
- Mohit Iyyer
- Kalpesh Krishna
- John Wieting
- Hendrik Strobelt
- Preslav Nakov
- Jure Leskovec
- Zhiyang Teng
Citations in the corpus
- GLTR: Statistical Detection and Visualization of Generated Text (2019, background): “Lavergne et al. (2008) introduce prediction entropy as an indicator of fake text.”
- Disinformation on the Web: Impact, Characteristics, and Detection of Wikipedia Hoaxes (2016, background): “Most of the work in this area has focused on engineering features that allow for detecting rumorous, fake, and deceptive content in social media [22].”
- Fast-DetectGPT: Efficient Zero-Shot Detection of Machine-Generated Text via Conditional Probability Curvature (2023, background): “These features encompass a range of measures, including relative entropy and perplexity (Lavergne et al., 2008), bag-of-words, average probability, and top-K buck-ets (Gehrmann et al., 2019), likelihood (Hashimoto et al., 2019; Solaiman et al., 2019), probability curvature (DetectGPT) (Mitchell et…”
- Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defense (2023, methodology): “Early methods detect statistical irregularities in entropy [Lavergne et al., 2008] and perplexity [Beresneva, 2016], while Gehrmann et al. [2019] introduced the GLTR visualizer to assist humans in detecting AI-generated text.”
- DetectLLM: Leveraging Log Rank Information for Zero-Shot Detection of Machine-Generated Text (2023, background): “Another stream of work attempts to distinguish machine-generated from human-written texts based on statistical irregularities in the entropy (Lavergne et al., 2008), perplexity (Beresneva, 2016) or in the n -gram frequencies (Badaskar et al., 2008).”