Optimizing Transfer Learning using Bidirectional Encoder Representations from Transformers

Natural language processing (NLP) is a branch of artificial intelligence that focuses on teaching computers how to analyze and understand human languages. The objective of this project is to develop advanced machine learning techniques to extract sentiment from text. We want to improve the current industry standards for sentiment analysis by leveraging the latest results from academic research. In particular, one of the techniques we will focus on is called Bidirectional Encoder Representations from Transformers (BERT). BERT is a technique that allows computers to learn more meaningful representation of language. By using techniques such as BERT, the company wants to teach computers how to better extract sentiment from text and use it to deliver superior service to customers.

Faculty Supervisor:

Yoshua Bengio

Student:

Partner:

Keatext - DUPLICATE

Discipline:

Computer science

Sector:

Information and Communications Technology; Technology

University:

Université de Montréal

Program: