Serdecznie zapraszamy 8 stycznia 2024 r. o godzinie 12:00 na Seminarium Centrum Marka Kaca pod tytułem Some applications of free random variables in cognitive sciences: Dynamics and entropy production in neural systems, które wygłosi prof. Maciej A. Nowak z IFT, UJ. Seminarium odbędzie się w sali D-2-43 Wydziału Fizyki, Astronomii i Informatyki Stosowanej UJ.
Abstract
The study of neuronal interactions is currently at the center of several big collaborative neuroscience projects (including the Human Connectome Project, the Blue Brain Project, the Brainome, etc.). The scientific targets range from attempts to obtain a detailed map of the entire brain to the developments of new, biologically inspired architectures of artificial neural networks. In the talk, I will give few examples, how free random variables can make an impact on these targets. I will start from the modeling of real neuronal systems – in particular I will show how well-known Rajan Abbott model can be rephrased and extended in the language of free random variables. I will stress the role of non-normality of neuronal adjacency matrices and vital role of overlaps of right and left eigenvectors, which turn out to be the main source of entropy production in non-equilibrium systems and complex transient dynamic behavior. Finally, provided time allows,
I conclude with few remarks, how free random variables can help to understand the black-box magic of deep neural networks.