“No one in my family understands what I do, so maybe this book will help! “, jokes physicist Julie Grollier, about her first book for children, Estelle and Noé discovering artificial intelligences (Millepages, 32 pages, 11.50 euros).
There is no doubt that the reader, young or old, will indeed grasp the subject, as the presentation is clear, without giving in to ease. The density of information is even quite impressive. In thirty boards drawn by Camouche, it is about the pioneer Alan Turing, binary language, the difference between memory and processor, the notion of algorithm (illustrated by a chocolate cake recipe, which Julie Grollier tested) and even ethics and neuroscience … In short, everything to initiate to this rising discipline. But ultimately little to understand what the researcher is really doing in this area!
Because Julie Grollier, 46, CNRS silver medalist in 2018, eminent member of the joint unit that this public body manages with the company Thales near Paris, is an artificial intelligence specialist of course, but in a branch a little apart.
An incomparable brain
If the stars of the field shine by the creativity or the power of their algorithms to play Go, drive autonomous cars, translate hundreds of languages, recognize speech …, the physicist is interested in the electronic components on which turn these calculations. There is an emergency because current systems consume a lot of electricity and will not be able to meet the requirements of algorithm designers or be embedded in mobile phones. “In terms of efficiency, we are 100 to 10,000 times behind our brain”, recalls the specialist, who is working for example on a European project, Ulpec, intended to develop an “ultra-low” consumption driving camera.
“I want to use physics to build a dynamic system that classes and learns, like the brain,” explains Julie Grollier
His specialty is to imagine new components, material analogues of those of our brain – namely not only the neurons, which transmit information, but also the synapses, which connect them and which are reinforced or attenuated according to the stimuli. “I want to use physics to build a dynamic system that classes and learns, like the brain. “ This is the neuromorphic pathway. “Julie is truly a pioneer in a whole area of neuromorphic calculus. She opened very original tracks there, recalls Albert Fert, 2007 Nobel Prize in Physics and Julie Grollier’s thesis director. After her defense in 2003 and two post-doc stays, she started this new idea in the lab. It was bloated. “
You have 67.03% of this article left to read. The rest is for subscribers only.
We want to give thanks to the writer of this write-up for this incredible web content
Julie Grollier, the physicist who creates artificial neurons