Deep Learning: Uncharted Territory in Sensors and Neural Networks

Explore deep learning's frontiers in this 25-minute conference talk from ETH's Prof. Luc Van Gool at INSAIT 2022. Delve into intriguing, uncharted territories of artificial intelligence, focusing on two key areas: the optimization of sensors in end-to-end training and the evolution of neural network architectures. Learn how current deep learning approaches differ from historical signal processing methods, and discover the potential for developing unified processing principles across multiple sensory modalities. Examine how the mammalian brain's neocortex architecture could inspire more integrated approaches to artificial intelligence, potentially unlocking new possibilities in processing various types of sensory inputs. Gain insights into how nature's solutions might guide the future development of more sophisticated and adaptable AI systems.