Le deep learning et ses applications en entreprise.

Abstract

The first concepts that led to what is now called deep learning are not new. They appeared in the fifties, when researchers tried to mimic the human brain behaviour. Since then, the history of what was called neural networks has experienced many ups and downs. Recent breakthroughs in their design combined with a bigger computing power and the availability of large data sets have made possible the use of deep learning algorithms to solve difficult problems in domains such as computer vision or language processing with a far better accuracy than with other existing methods.

The purpose of this master thesis is to study the true potential of deep learning applications in the real world with a special focus on the impacts that they can have on the world of work.

While there is still a lot of research ongoing and that many applications will not be ready before many years, it appears that companies can already use deep learning in many ways but that it does not always lead to the same result. Some companies are engaged in a race involving a lot of investment to be the leaders of a revolution. Others try to benefit from the known advantages of the technology to offer new services or alternatives to existing ones without having such a revolutionary impact on the world.

Massive job losses are to be expected with the deployment of some of those applications and the others will definitely have an impact on the daily job of many workers. Nevertheless, in the short and medium term, this impact will most probably be to help them with new tools and to free them from the most repetitive tasks while leaving them the orchestration and the important decisions. It will therefore result to a better quality of services and products in many domains.