Zoom-talk: Machine Learning for Variational Quantum Algorithms

Speaker: Viroshaan Uthayamoorthy

Title: Machine Learning for Variational Quantum Algorithms

Abstract: The training of the parameters of variational quantum algorithms (VQAs) is in general very challenging. We will discuss different parameter initialization heuristics and a machine learning technique to learn the patterns in optimal QAOA parameters. Using this approach, one can reduce the number of intermediary optimization runs needed to find good local optima. Relatively few graph instances are needed to learn the optimal QAOA parameters using a feedforward neural network. In addition, we will present an approach that uses a classical neural network to escape potential “bad” local minima. All procedures were tested on a suite of examples for the well-known MaxCut problem. The results are based on a master thesis.

Join Zoom Meeting: https://zoom.us/j/92762737127