A machine learning approach to improve scalability and robustness of variational quantum circuits

Abstract

Quantum computing is an emerging new field that aims to leverage the power of a “quantum computer” to solve problems which are currently considered to NP-Hard or NP-Complete. The key idea is to encode inputs as quantum states and device a system where the measured outcomes correspond to a solution of the given problem. While a fault-tolerant quantum computer is still a theoretical possibility, we are able to evaluate the potential of quantum algorithms by running them on a class of devices called Noisy Intermediate Scale Quantum (NISQ) computers. ☐ The advantage of having access to NISQ computers is that they allow for immediate verification of the speedup provided by a proposed quantum algorithm. However, there are significant downsides to the current generation of these devices. Most notable of them are limited gate depth, high sensitivity to noise, ability to scale to only a few number of qubits and a tendency to get stuck in “barren plateaus”. In this research proposal, we introduce and define some key problems encountered in the simulation of quantum algorithms on NISQ devices. We then propose mitigating solutions inspired from machine learning methods and show the ecacy of our methods by simulating different types of variational algorithms on NISQ devices.