Discover Deep Learning

Written by Perrine Croatto, on 09 April 2019

Deep Learning, what are we talking about?

Deep Learning refers to the automatic learning algorithm family that enables the machine to learn from their data. Although these algorithms were discovered during the 1950’s, the computing power has never been powerful enough to use them. Today we use GPU cards (Graphic Processing Unit). Even though, these cards were initially designed to be used by gamers. We are now able to use ultra-fast calculations and to make the training of these algorithms possible. For the past ten years, databases have gained substantial volume of data. With the development of IoT, social media, open data, etc., we now have infinite possibilities with Deep Learning.

These data are necessary to run calculation models, they could be qualified as the raw material of Deep Learning! Generally speaking, the more data there are, the more performant the models become. Deep Learning models allow numerous recent technologies to perform well (and sometimes overcome humans!), which is the case of face recognition, vocal recognition, computer vision or language treatment. Lastly an important buzz came out around AI and Deep Learning thanks to the enthusiasm shown by researchers in this domain and to new models that revolutionized several fields.

To sum up: Deep Learning = Data + Model + Calculation power

What does Aneo regarding this topic?

Essentially, Deep Learning is a very recent field which is starting to become important in the industrial domain. The companies that are starting to understand its interest and its power are more and more willing to test it. Many specialized masters about data science, machine learning or deep learning have been created… Numerous tools already exist to facilitate the creation of models, allowing them to be designed easily and several news or blog articles deal with this subject.

Today, at Aneo, as a data consulting firm, we are starting to specialize ourselves in DL. It is of course a highly technical domain. But the door is left wide open and I guess that we are on the right way.

On which domain are we working on at Aneo, and on which of them are we particularly performant?

We are working on making people understand the interests of Deep Learning. Indeed, the questions it raises, tend to appear more important for our clients. Also, we work on a new and innovating field, it is important to adapt ourselves to it. Aldwin has the expertise to succeed in this domain.

Aneo also owns a strong Data Science culture, which will help us in this process. Moreover, Aneo also rhyme with IT, data, experience with customers and associates, agility as well as optimization-performance.

How do you see Deep Learning evolve in the upcoming years?

This field generates great enthusiasm and it is an active research field. As a result, there is a strong media hype regarding Artificial Intelligence focusing more specifically on Deep Learning; the interest either comes from the experts’ community or the general public.

This has led to an ever-growing number of publications. But I think it can only increase, the only question is when the explosion will happen.