SotA on Deep Learning

By ITEA Vice-chairman Philippe Letellier

Artificial Intelligence is the new wave of our digital society and it is visible in a large majority of the ITEA projects. Nowadays, we are already producing a huge amount of data with all of our products/services. The next challenge is thus to extract from this mountain of data, some new understanding of user usage and needs to enhance our products/services even more so as to actually strengthen their customer orientation. It is a typical ITEA target. A key tool in this direction is what we call Deep Learning.

In our ITEA living roadmap (ITEA account and NDA required) we have many SotAs on this Deep Learning topic. Among these I invite you to have a look at the "SOMEDI 2.2.3. DEEP LEARNING" document. It is a short State of the Art on two kinds of interesting neural network algorithms: Recurrent Neural Networks and Long Short-Term Memory. It also describes a set of open source tools for this deep learning approach.

Another document I recommend is the "PSCrimson 2.2. SotA overview", which is more focused on the Convolutional Neural Networks algorithm family for computer vision.

However, the best advice I can give you is to regularly browse the ITEA living roadmap where you will find many interesting SotAs on Deep Learning and more in general on all the domains of Artificial Intelligence as already worked on by so many ITEA projects.

Have a good read of these SotAs and exploit them well to push new innovation in this promising Artificial Intelligence direction.