Optimizing meta-parameters for quantum machine learning

Variational quantum circuits allow to do machine learning on near-term quantum hardware. The success of the algorithms strongly depends on the choices made during the model design. These choices can be quantified and are called meta-parameters. Examples are the number and type of quantum gates in the circuit, the number of layers of circuit template, the choice of training data, the optimization algorithm, and the optimization strategy. Francisco will be studying these in the context of the compression and classification of quantum data and will try to propose a best practice. This research fits in a larger project about detecting quantum phase transitions in spin chains. The expected outcomes are twofold: we will advance the knowledge of efficent algorithms in quantum machine learning and we will gain understanding of the quantum phases encoded in the wave function of our test system.

Faculty Supervisor:

Alán Aspuru-Guzik

Student:

Partner:

The University of Texas at Austin

Discipline:

Physics

Sector:

Education

University:

University of Toronto

Program: