講演情報
[O1-5]How genes and brains made mathematics?
○鎌谷 直之 (株式会社スタージェン 医療人工知能研究所)
Three important systems, genes, the brain, and artificial intelligence have similar goals, namely, the maximization of likelihood or minimization of cross-entropy. Animal brains have evolved through predator-prey interactions in which maximizing probabilities of survival and transmission of genes to offspring were the main objectives. I propose that mathematics is a structure of thinking developed through reinforcement learning in the brain to process data from sensory organs (input) and to move muscles (output) and thereby win predator-prey fights. I will explain how various structures of mathematics including sets of numbers, algebraic structures, differentiation/integration, trigonometric functions and even Eigen value/vector calculation evolved through the predictor-prey fights. Importantly, the brains are made by genes, and only the latter are transmitted through generations. The main difference between gene and brain systems as learning systems is the interval of the parameter updates. This difference renders the brain the ability to control much larger numbers of parameters than genes and use non-linear models. I predicted that the future frontier of science would be about the cognitive processes that scientists’ brains use to study mathematics and physics and how the genes control such processes (Kamatani N. JHG 2020). Papers have begun to appear to show that my prediction is correct. Thus, deep learning has started to guide mathematicians’ intuition to discover new conjectures and theorems and help physicists to find accurate solutions to Schroedinger equation.