From biological to artificial intelligence and back again
A talk from dr. Mihai Petrovici, researcher in the Human Brain Project from the Computational Neuroscience Group of the Department of Physiology, University of Bern
Friday, December 13th, 2019, 19:00 | CINETic – Multifunctional Hall (3B, Tudor Arghezi street, entrance through the TNB parking lot)
Free entry | Please register by filling in this FORM
Ever since its golden age in the 50s and 60s of the last century, science fiction has taught us that, sooner rather than later, humanity will have to deal with artificial intelligence (with more or less apocalyptic consequences). Early pioneers in AI research, such as Marvin Minski or Herbert Simon, have also done their utmost to nourish these expectations. However, the year is 2019 and Terminators are nowhere in sight. This is certainly reassuring, but also somewhat peculiar, given that every one of us carries with them, in their pocket, a much more capable computer than the one that put the Apollo astronauts on the moon.
Our brain remains light-years ahead of the most powerful supercomputers of our time, while requiring only the energy equivalent of a couple of bananas per day in order to function properly. Today, we know that this impressive feat is mainly due to the neocortex, a complex network of tens of billions of neurons, whose dynamics differ radically from modern computer architectures. Can we understand its underlying principles of computation? And can we maybe even cast them directly into silicon?
Today, we can attempt a cautiously optimistic answer to both these questions. What started out as a vague idea only a few decades ago is now becoming reality: modeled on the conceptual scaffold of cortical circuitry, new so-called neuromorphic algorithms and architectures are being developed. In this talk, we will explore some of these concepts and observe these machines as they learn to think on their own. And last but not least, we will also take a peek into their dreams.
Dr. Mihai Petrovici
My main area of research is bio-inspired AI, with a particular focus on ensemble phenomena in neural networks, Bayesian inference with spikes, learning in hierarchical networks and the development of beyond-von-Neumann architectures capable of embedding functional neural network models.
Currently, my home base is the Computational Neuroscience Group at the Department of Physiology, University of Bern, which I am co-leading with Prof. Walter Senn. I am also the founder and current leader of the theory and modeling department of the Vision(s) group in Heidelberg. I believe that there is much to learn from biology about cognition, but I am more of a functionalist when it comes down to actually building physical implementations – there are good reasons for airplanes not to flap their wings. In our groups, we therefore combine knowledge and methods from neuroscience, information geometry, the physics of classical complex systems, machine learning and microelectronics to design functional and robust neuronal network models and embed them into low-power, highly accelerated neuromorphic devices.