Model Attribution in LLM-Generated Disinformation: A Domain Generalization Approach with Supervised Contrastive Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Beigi, Alimohammad, Tan, Zhen, Mudiam, Nivedh, Chen, Canyu, Shu, Kai, Liu, Huan
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913467681734656
author Beigi, Alimohammad
Tan, Zhen
Mudiam, Nivedh
Chen, Canyu
Shu, Kai
Liu, Huan
author_facet Beigi, Alimohammad
Tan, Zhen
Mudiam, Nivedh
Chen, Canyu
Shu, Kai
Liu, Huan
contents Model attribution for LLM-generated disinformation poses a significant challenge in understanding its origins and mitigating its spread. This task is especially challenging because modern large language models (LLMs) produce disinformation with human-like quality. Additionally, the diversity in prompting methods used to generate disinformation complicates accurate source attribution. These methods introduce domain-specific features that can mask the fundamental characteristics of the models. In this paper, we introduce the concept of model attribution as a domain generalization problem, where each prompting method represents a unique domain. We argue that an effective attribution model must be invariant to these domain-specific features. It should also be proficient in identifying the originating models across all scenarios, reflecting real-world detection challenges. To address this, we introduce a novel approach based on Supervised Contrastive Learning. This method is designed to enhance the model's robustness to variations in prompts and focuses on distinguishing between different source LLMs. We evaluate our model through rigorous experiments involving three common prompting methods: ``open-ended'', ``rewriting'', and ``paraphrasing'', and three advanced LLMs: ``llama 2'', ``chatgpt'', and ``vicuna''. Our results demonstrate the effectiveness of our approach in model attribution tasks, achieving state-of-the-art performance across diverse and unseen datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2407_21264
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Model Attribution in LLM-Generated Disinformation: A Domain Generalization Approach with Supervised Contrastive Learning
Beigi, Alimohammad
Tan, Zhen
Mudiam, Nivedh
Chen, Canyu
Shu, Kai
Liu, Huan
Computation and Language
Model attribution for LLM-generated disinformation poses a significant challenge in understanding its origins and mitigating its spread. This task is especially challenging because modern large language models (LLMs) produce disinformation with human-like quality. Additionally, the diversity in prompting methods used to generate disinformation complicates accurate source attribution. These methods introduce domain-specific features that can mask the fundamental characteristics of the models. In this paper, we introduce the concept of model attribution as a domain generalization problem, where each prompting method represents a unique domain. We argue that an effective attribution model must be invariant to these domain-specific features. It should also be proficient in identifying the originating models across all scenarios, reflecting real-world detection challenges. To address this, we introduce a novel approach based on Supervised Contrastive Learning. This method is designed to enhance the model's robustness to variations in prompts and focuses on distinguishing between different source LLMs. We evaluate our model through rigorous experiments involving three common prompting methods: ``open-ended'', ``rewriting'', and ``paraphrasing'', and three advanced LLMs: ``llama 2'', ``chatgpt'', and ``vicuna''. Our results demonstrate the effectiveness of our approach in model attribution tasks, achieving state-of-the-art performance across diverse and unseen datasets.
title Model Attribution in LLM-Generated Disinformation: A Domain Generalization Approach with Supervised Contrastive Learning
topic Computation and Language
url https://arxiv.org/abs/2407.21264