ICNLSP 2024: Dual-Task Learning for AI-Generated Medical Text Detection and Named Entity Recognition
Автор: ICNLSP Conference
Загружено: 2024-12-11
Просмотров: 24
Описание:
Dual-Task Learning for AI-Generated Medical Text Detection and Named Entity Recognition
By: Saja B. Al-Dabet, Ban Alomar, Sherzod R Turaev, Abdelkader Belkacem
United Arab Emirates University
7th International Conference on Natural Language and Speech Processing.
https://icnlsp.org/2024welcome
Abstract:
"The integration of artificial intelligence (AI) into the medical field has revolutionized documentation and diagnosis. However, the de tection of AI-generated text within medical
records remains a crucial task. This paper describes a dual-task learning framework using the ELECTRA model for detecting AI-generated medical texts and performing named entity recognition (NER). The dual-task model includes a binary classification head for identifying AI-generated texts and an NER head for extracting medical entities. Experiments on radiology report and medical texts datasets show that the proposed approach achieves robust performance, with F1 scores of 0.985 and 0.996 for classification and 0.51 and 0.68 for NER. The model achieves a high accuracy of 0.996 for medical text classification and 0.985 for MiMic classification, enhancing automated medical text analysis and supporting clinical decision-making."
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке:
