Hateful Meme Detection through Context-Sensitive Prompting and Fine-Grained Labeling

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Ouyang, Rongxin, Jaidka, Kokil, Mukerjee, Subhayan, Cui, Guangyu
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909392447733760
author Ouyang, Rongxin
Jaidka, Kokil
Mukerjee, Subhayan
Cui, Guangyu
author_facet Ouyang, Rongxin
Jaidka, Kokil
Mukerjee, Subhayan
Cui, Guangyu
contents The prevalence of multi-modal content on social media complicates automated moderation strategies. This calls for an enhancement in multi-modal classification and a deeper understanding of understated meanings in images and memes. Although previous efforts have aimed at improving model performance through fine-tuning, few have explored an end-to-end optimization pipeline that accounts for modalities, prompting, labeling, and fine-tuning. In this study, we propose an end-to-end conceptual framework for model optimization in complex tasks. Experiments support the efficacy of this traditional yet novel framework, achieving the highest accuracy and AUROC. Ablation experiments demonstrate that isolated optimizations are not ineffective on their own.
format Preprint
id arxiv_https___arxiv_org_abs_2411_10480
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Hateful Meme Detection through Context-Sensitive Prompting and Fine-Grained Labeling
Ouyang, Rongxin
Jaidka, Kokil
Mukerjee, Subhayan
Cui, Guangyu
Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
Machine Learning
Multimedia
68T45, 68T50, 68T07
I.2.10; I.2.7; I.2.6
The prevalence of multi-modal content on social media complicates automated moderation strategies. This calls for an enhancement in multi-modal classification and a deeper understanding of understated meanings in images and memes. Although previous efforts have aimed at improving model performance through fine-tuning, few have explored an end-to-end optimization pipeline that accounts for modalities, prompting, labeling, and fine-tuning. In this study, we propose an end-to-end conceptual framework for model optimization in complex tasks. Experiments support the efficacy of this traditional yet novel framework, achieving the highest accuracy and AUROC. Ablation experiments demonstrate that isolated optimizations are not ineffective on their own.
title Hateful Meme Detection through Context-Sensitive Prompting and Fine-Grained Labeling
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Computation and Language
Machine Learning
Multimedia
68T45, 68T50, 68T07
I.2.10; I.2.7; I.2.6
url https://arxiv.org/abs/2411.10480