ML for Quantum Error Mitigation
Efficient and scalable QEM using machine learning
This paper investigates the practical utility of machine learning for quantum error mitigation (ML-QEM) which primarily has the advantage of being arguably the most efficient QEM method at runtime—we ask the following questions:
a) How well does it perform under realistic noise?
b) How efficient it can be even if considering training overhead?
c) In what applications can it be particularly useful?
d) Whether it be scaled to classically intractable circuits?
Our work suggests positive results in all these aspects.
I had the opportunity to discuss this at the live-streamed weekly Qiskit Seminar on Qiskit’s YouTube channel, and at the IEEE Quantum Week 2023.
The research code for all the simulations and experiments can be found at this Github repo.