Yesmina JAAFA, a doctoral student at SEGULA/iCube Laboratory has been selected to present her researches at the “The International Conference on Machine Learning” (ICML 2019): (https://icml.cc/)

It is one of the most prestigious conferences of the world in the Artificial Intelligence field and will regroup hundreds of world-renowned industrial and academic affiliates. It is organized by a committee of top researchers from Microsoft Research, Facebook AI Research, Google Brain, DeepMind and universities such as Berkeley, Cornell and Carnegie Mellon. The number of items proposed is very high (over 3000) and being selected is remarkable.

Our article “Meta-Reinforcement Learning for Adaptive Autonomous Driving” focuses on a learning algorithm applied to autonomous vehicles.

To better understand:

Current IA are able to learn in particular situations but often have difficulties generalizing their learning to variable and non-stationary environments.

For example, the algorithm learns to drive on an urban circuit, on a sunny day and on a dry road. But it will not be effective on the same circuit if it rains or if the day drops and will have to relearn to drive. Yesmina’s researches propose an approach that enables the acquisition and transfer of knowledge to several application contexts through meta-learning mechanisms or “learning to learn”. Her work has been developed and tested on an ultra-realistic simulator, CARLA.

Read the entire Yesmina’s article on this link: https://openreview.net/pdf?id=S1eoN9rsnN

This article is about a cutting-edge topic that our customers have been asking for. It allows us to have a positive image that helps us position ourselves on the new Autonomous Vehicle business.

So, Yesmina will be there until June 15th at Long Beach in California in order to present her studies to scientists and industrialist persons.

Congratulations and good luck to her!