Word Embeddings and Ethics in NLP
Автор: ICT4DEV
Загружено: 2023-07-12
Просмотров: 109
Описание: Natural language models are becoming more and more popular thanks to their semantic accuracy and flexibility. In fact, word and sentence embeddings are used to enhance a variety of applications, from CV parsing to web search. However, concern is rapidly growing around their display of social bias and lack of transparency: in particular, the amount of gender bias risks favouring the consolidation and amplification of social stereotypes. In this talk, Tommaso Dolci, a Research Assistant at Politecnico di Milano, presents a brief introduction to word and sentence embeddings and discusses different methodologies to measure and mitigate gender bias in these models.
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке:
![Почему реактивный двигатель не плавится? [Veritasium]](https://imager.clipsaver.ru/F6UB0V2Fct4/max.jpg)