Description

The recent advances in Neural Machine Translation have the potential to revolutionise the way the world communicates and to eliminate language barriers. One of the main issues for any deep learning technique is the availability of large amounts of data, which typically requires human intervention to generate.

About this lecture

Vasco Pedro will explore the recent advances in neural machine translation, the hybrid model of human data generation, and the potential to reshape the world of multilingual communication.

Furthermore, he will talk about Unbabel and how their team worked on developing a scalable model of human/machine interaction. The focus is on language translation where the machine augments the human and the human augments the machine. Pedro will also present the result of this effort: a highly scalable data generation machine that could significantly increase the speed with which we can train neural machine translation models.

We value your feedback and welcome any comments you may have to help us improve our programs. Make sure you submit the survey after you have attended this event!