Machine learning is one of the fields that could benefit from near-term quantum computers: just the same way massively parallel digital computers enabled deep learning to scale up, quantum processing units (QPUs) are great at doing certain workloads. The problem is that the emergent field of quantum machine learning has been plagued with expectations that are unrealistic on contemporary quantum computers and relevance to the machine learning and AI communities has largely been overlooked.

In this talk, we give a survey on what early quantum devices can contribute to machine learning. The primary algorithmic primitives are sampling, optimization, calculating kernel functions, and some variational problems efficiently which map to hybrid classical-quantum protocols. The main application areas in machine learning are probabilistic graphical models, in particular Boltzmann machines and deep variants thereof, quantum neural networks, and searches over discrete parameter spaces. These models have different strengths than the ones trained on digital computers, hence quantum machine learning plays a complementary role to classical techniques, rather than acting as a replacement. We will also highlight possible pathways forward that would make upcoming quantum architectures more relevant to AI research.   

Peter Wittek is an Assistant Professor in the University of Toronto and an affiliate in the Vector Institute for Artificial Intelligence the Perimeter Institute for Theoretical Physics. He obtained his PhD from the National University of Singapore. His research explores the synergies between artificial intelligence, machine learning, quantum information theory, and quantum computing.  He authored the first book on quantum machine learning and launched the first online course on the subject on edX. As the Academic Director of the Quantum Program in the Creative Destruction Lab, he oversees two dozen quantum software startups a year that exploit contemporary quantum technologies in a commercial sett