The microprocessors and graphics processing units that implement artificial intelligence compute and store the information in different units, and a lot of time and energy is wasted on data transfer.

In this talk, I will present a radically new computing paradigm called neuromorphic computing, which in small systems has shown to be 1000 times more energy efficient. This new approach is based on the deployment of electronic neurons and synapses for in-memory computation, and I will show you how our team has reduced their complexity, opening the door for the upscaling and commercialization of energy efficient neuromorphic computing systems for artificial intelligence.

www.lanzalab.org | LinkedIn

Mario Lanza got his PhD in Electronic Engineering in 2020 at the Autonomous University of Barcelona. He was Marie Curie postdoctoral Fellow at Stanford University (USA), and lead research groups in Soochow University (China, 2013-2020) and King Abdullah University of Science and Technology (Saudi Arabia, 2020-2024).

Currently, he is Professor at National University of Singapore, where he works on the development of energy-efficient hardware for artificial intelligence. He has published multiple research articles in top journals like Nature and Science, and has been distinguished as Fellow by the Institute of Electrical and Electronic Engineers (IEEE) of the USA.