Google Tech Talks
December, 13 2007
This tech talk series explores the enormous opportunities afforded by the emerging field of quantum computing. The exploitation of quantum phenomena not only offers tremendous speed-ups for important algorithms but may also prove key to achieving genuine synthetic intelligence. We argue that understanding higher brain function requires references to quantum mechanics as well. These talks look at the topic of quantum computing from mathematical, engineering and neurobiological perspectives, and we attempt to present the material so that the base concepts can be understood by listeners with no background in quantum physics.
In this second talk, we make the case that machine learning and pattern recognition are problem domains well-suited to be handled by quantum routines. We introduce the adiabatic model of quantum computing and discuss how it deals more favorably with decoherence than the gate model. Adiabatic quantum computing can be understood as an annealing process that outperforms classical approaches to optimization by taking advantage of quantum tunneling. We also discuss the only large-scale adiabatic quantum hardware that exists today, built by D-Wave. We present detailed theoretical and experimental evidence showing that the D-Wave chip does indeed operate in a quantum regime. We report about an object recognition system we designed using the adiabatic quantum computer. Our system uses a combination of processing steps, where some are executed on classical hardware while others take advantage of the quantum chip. Both interest point selection and feature extraction are accomplished using classical filter operations reminiscent of receptive field properties of neurons in the early visual pathways. Image matching then proceeds by maximizing geometrical consistency and similarity between corresponding feature points, which is an NP-hard optimization problem. To obtain good solutions, we map this to the problem of finding the minimum energy in an Ising model in which the vertices represent candidate match pairs, bias terms reflect feature similarity, and interaction terms account for geometric consistency. The adiabatic quantum computer is then employed to find a low energetic minimum of the Ising dynamics. We conclude with a look towards which type of machine learning problems maybe most suitable for mapping to a quantum computing architecture.
Speaker: Hartmut Neven
Speaker: Dr. Geordie Rose
Geordie Rose is a founder and CTO of D-Wave. He is known as a leading advocate for quantum computing and physics-based processor design, and has been invited to speak on these topics in venues ranging from the 2003 TED Conference to Supercomputing 2005.
His innovative and ambitious approach to building quantum computing technology has received coverage in BC Business, The Vancouver Sun, Vancouver magazine, The Globe and Mail, The National Post, USA Today, MIT Technology Review magazine, the Harvard Business Review and Business 2.0 magazine, and one of his business strategies was profiled in a Harvard Business School case study. He has received several awards and accolades for his work with D-Wave, including being short-listed for a 2005 World Technology Award.
Dr. Rose holds a PhD in theoretical physics from the University of British Columbia, specializing in quantum effects in materials. While at McMaster University, he graduated first in his class with a BEng in Engineering Physics, specializing in semiconductor engineering.
Since the inception of D-Wave in 1999, Dr. Rose, as founding CEO, raised over $45M on behalf of the company, including a round led by Draper Fisher Jurvetson (DFJ) — the first ever investment by a top-tier US venture capital firm in quantum computing.