Maria Kieferova - Training quantum neural networks with an unbounded loss function - IPAM at UCLA

Recorded 27 January 2022. Maria Kieferova of the University of Technology Sydney presents “Training quantum neural networks with an unbounded loss function“ at IPAM's Quantum Numerical Linear Algebra Workshop. Abstract: Quantum neural networks (QNNs) are a framework for creating quantum algorithms that promises to combine the speedups of quantum computation with the widespread successes of machine learning. A major challenge in QNN development is a concentration of measure phenomenon known as a barren plateau that leads to exponentially small gradients for a range of QNNs models. In this work, we examine the assumptions that give rise to barren plateaus and show that an unbounded loss function can circumvent the existing no-go results. We propose a training algorithm that minimizes the maximal Renyi divergence of order two and present techniques for gradient computation. We compute the closed form of the gradients for Unitary QNNs and Quantum Boltzmann Machines and provide sufficient conditio