A lot of folks are researching using quantum computing to work with artificial intelligence algorithms. Recently, IBM and Intel received some attention (primarily because they have larger PR departments than most others) for papers they recently published.
IBM published a paper showing how quantum computing can be used on a supervised learning algorithm to classify data using a technique called feature mapping. The distinguishing feature to us in their approach was that it was designed to use short depth circuits which is preferred in the current era of Noisy Intermediate Scale Quantum (NISQ) computers.
Short-depth refers to the fact that a qubit only goes through a small number of sequential gate operations before it is measured and put into a classical 0 or 1 status. This is important when one is working with quantum gates that only have fidelities in the 99% range. It doesn’t take very many sequential gate operations before the error amount obscures any valid states that would have been computed by the algorithm. Not all Quantum Machine Learning (QML) algorithms use this approach and they would require quantum computers with much higher levels of qubit quality which will not be available in the near term.
The specific IBM example cited in the paper only required two qubits and two qubit operations can be easily simulated on any classical computer. But we expect IBM to continue this research for more complex problems that will be more challenging for a classical computer.
Intel’s paper is a little unique because instead of using quantum computing to enhance AI, they are using AI to better understand quantum physics phenomena. Their paper describes using the latest advancement in deep neural networks to simulate faster and more thoroughly the smallest of particles and how they interact at the quantum level.
For more information, you can view IBM’s Nature paper on supervised learning here and the associated research blog here. You can also view Intel’s paper in Physical Review Letters here and the associated news release here.