Quantum physical machine learning

26^th March 2025, 14:00 Heleshaw Lecture Theatre, School of Engineering
Gerard Milburn
University of Queensland

Abstract

Why does AI consume so much energy? What are the physical constraints on machines that learn? I will discuss the thermodynamics of learning machines with a particular emphasis on quantum constraints in physical neural networks and kernel machines. A careful examination of experimental implementations using superconducting circuits and nano electronics illustrate the physical principles. Quite apart from a thermodynamic advantage, does quantum offer any computational advantage? Current indications suggest that while quantum offers energy efficiency it may not give a computational advantage. I will offer a more optimistic assessment based on recent work in quantum chaos.

Biography

Gerard Milburn obtained a PhD in theoretical Physics from the University of Waikato in 1982 for work on squeezed states of light and quantum non demolition measurements. In 1994 he was appointed Professor in Physics at The University of Queensland. In 2011 he became the founding Director of the Australian Research Council Center of Excellence for Engineered Quantum Systems. He is currently lecturer in the School of Mathematics and Physics at Sussex University, and Quantum Fellow at the National Quantum Computing Centre at Rutherford National Laboratory.

He has worked in the fields of quantum optics, quantum stochastic processes, atom optics, quantum control, engineered quantum systems. In 2001, together with Knill and Laflamme he published a scheme for quantum computing with photons, known as the KLM scheme, that had a major impact on the experiments in quantum optics. His current research focuses on the physical constraints in autonomous learning machines.

Gerard Milburn is a Fellow of the Royal Society of London, Fellow of the Australian Academy of Science and the American Physical Society. He is the author of five books including Quantum Optics (with Dan Walls), Quantum Measurement and Control (with Howard Wiseman) and Quantum Optomechanics (with Warwick Bowen).