Neuromorphic Computing – From Neuroscience Inspiration to AI Applications
Speaker: Prof. Rainer Waser
Aachen University, Germany
DEIB - Conference Room "E. Gatti" (Bld. 20)
October 17th, 2024 | 11.00 am
Contact: Prof. Daniele Ielmini
Aachen University, Germany
DEIB - Conference Room "E. Gatti" (Bld. 20)
October 17th, 2024 | 11.00 am
Contact: Prof. Daniele Ielmini
Sommario
On October 17th, 2024 at 11.00 am the seminar titled "Neuromorphic Computing – From Neuroscience Inspiration to AI Applications" will take place at DEIB Conference Room "Emilio Gatti" (Building 20).
In information technology, there is currently a transition from classical instruction-based computation, where algorithms are the recipes for data processing and lead to predictable results, to AI-based computation, which relies on machine learning and the automatic extraction of knowledge from data to produce results with a certain probability. The clear downside of this development is the enormous and increasing energy consumption of computing systems. For this reason, I will briefly motivate my talk with the climate change.
In our projects, we are investigating devices and systems for Neuromorphic Computing (NC) that are more energy-efficient than classical approaches. The focus of our research is on redox-based memristive elements. In this talk, fundamental aspects of the physics of memristive elements for NC will be outlined. Material physics will be linked to neighboring disciplines, neuroscience and computer science / artificial intelligence (AI). It will be shown that such an interdisciplinary approach brings the highest benefit in a future energy-efficient NC. In the outlook, we consider the implications of memristive-based NC and its potential impact on future AI. This will lead to aspects of advantages and threats by AI on a personal and societal level.
In information technology, there is currently a transition from classical instruction-based computation, where algorithms are the recipes for data processing and lead to predictable results, to AI-based computation, which relies on machine learning and the automatic extraction of knowledge from data to produce results with a certain probability. The clear downside of this development is the enormous and increasing energy consumption of computing systems. For this reason, I will briefly motivate my talk with the climate change.
In our projects, we are investigating devices and systems for Neuromorphic Computing (NC) that are more energy-efficient than classical approaches. The focus of our research is on redox-based memristive elements. In this talk, fundamental aspects of the physics of memristive elements for NC will be outlined. Material physics will be linked to neighboring disciplines, neuroscience and computer science / artificial intelligence (AI). It will be shown that such an interdisciplinary approach brings the highest benefit in a future energy-efficient NC. In the outlook, we consider the implications of memristive-based NC and its potential impact on future AI. This will lead to aspects of advantages and threats by AI on a personal and societal level.
Biografia
Rainer Waser is Professor of Electrical Engineering at RWTH Aachen University. He is also director of the section Electronic Materials at the Peter Grünberg Institute which is located on the campus of Jülich Research Center (Forschungszentrum Jülich). His research and teaching is on solid-state chemistry and defect chemistry to electronic properties and modelling, the technology of new materials and the physical properties of construction components. Important findings include insights in the functioning of the so-called memristors. He studied Physical Chemistry at Darmstadt University of Technology where he received a diploma degree in 1979. Then he went to the University of Southampton to conduct research at the Institute of Electrochemistry. After that he turned to Darmstadt and worked as scientific assistant until he completed his PhD. Waser joined the Philips research laboratories (research group Electronic Ceramics) at Aachen. In 1992, Waser accepted a Chair for Electronic Materials in the Faculty of Electrical Science and Information Technology at RWTH Aachen University. In 2012, Waser was elected to the post of Speaker of the Department of Electrical Engineering and Information Technology at Aachen university. Waser was awarded the renowned Gottfried Wilhelm Leibniz Prize in 2014.
Rainer Waser (waser@iwe.rwth-aachen.de)
Rainer Waser (waser@iwe.rwth-aachen.de)