A path towards Quantum Advantage in training Deep Generative Models with Quantum annealers

A path towards Quantum Advantage in training Deep Generative Models with Quantum annealers Illustration of a Restricted Boltzmann Machine (RBM) bipartite graph where viviv_i are visible nodes, hjhjh_j are hidden nodes and wijwijw_{ij} are the weights connecting the hidden and visible nodes.

A team of researchers from D-Wave, NASA Ames Research Center and University of Florence has successfully employed D-Wave quantum annealers as Boltzmann samplers to perform quantum-assisted, end-to-end training of QVAE (Quantum Variational AutoEncoder).

The development of quantum-classical hybrid (QCH) algorithms is critical to achieve state-of-the-art computational models. A QCH variational autoencoder (QVAE) consists of a classical auto-encoding structure realized by traditional deep neural networks to perform inference to, and generation from, a discrete latent space. The latent generative process is formalized as thermal sampling from either a quantum or classical Boltzmann machine (QBM or BM). This setup allows quantum-assisted training of deep generative models by physically simulating the generative process with quantum annealers.

The hybrid structure of QVAE allows to deploy current-generation quantum annealers in QCH generative models to achieve competitive performance on datasets such as MNIST. The results suggest that commercially available quantum annealers can be deployed, in conjunction with well-crafted classical deep neutral networks, to achieve competitive results in unsupervised and semisupervised tasks on large-scale datasets. The team also provide evidence that their setup is able to exploit large latent-space (Q)BMs, which develop slowly mixing modes. This expressive latent space results in slow and inefficient classical sampling, and paves the way to achieve quantum advantage with quantum annealing in realistic sampling applications. (SciRate)

Read more.