Strong Generalization From Small Brains and No Training Data - Neural Computations in Complex Environments

Watch a 58-minute lecture from UC Berkeley's Bruno Olshausen at the Simons Institute exploring how animals develop geometric representations of their environment for survival and future adaptation. Discover recent advances in understanding neural computations that enable robust operation with minimal power consumption and compact form factors. Learn about the mechanisms behind strong generalization capabilities in biological systems that function effectively without extensive training data, offering insights into how small brains achieve complex environmental navigation and adaptation to unknown future scenarios.