Why is fake news so hard to catch and how can we fight it? - Naver Labs Europe
loader image

meetup logoData Science Grenoble, meetup:  Tuesday, 15th October 2019, 7:00 pm

Speaker: Julien Perez

Recent, dramatic improvements in natural language understanding and generation are partially due to advances in machine learning and, in particular, deep learning. In association with this progress, research in text-based question-answering (machine-reading) and neural generation models, have given rise to many opportunities in the domains of information retrieval, generation, translation and summarization. However, these recent developments have also led to the proliferation of ‘fake news’. Yet, whilst we start to grapple with the problem of fake news created and published by humans, we already have a new challenge in the shape of “autonomous fake news generation” whereby models can automatically generate fake information and on a very large scale. In this presentation, we will talk about the challenges in detecting ‘fake news’ before detailing the current neural approaches of machine writing based on parametric language modelling that actually enable the automatic generation of fake news.
We’ll conclude by presenting the principles that lie behind the main countermeasures that are currently being pursued.

This event will take place at Cowork in Grenoble.



Ceci correspond à une petite biographie d'environ 200 caractéres