Implementation of Quantum Restricted Boltzmann Machine using quantum annealing on D-wave's QPU.
Publication available at: https://cmst.eu/articles/applying-a-quantum-annealing-based-restricted-boltzmann-machine-for-mnist-handwritten-digit-classification/
In collaboration with Mateusz Slysz
NOTE: previous repository utilizing IBM's QPU moved to QRBM-qiskit
To run demo, click on the badge below:
To download the default dataset, click here: MNIST - a dataset of pictures of handwritten digits with 28x28 pixel resolution
The best way to run this code is to download the dataset and run all cells in the jupyter notebook.
A plot showing how the generated digit zero changes for different number of digits trained by the quantum RBM with respect to the chain strength parameter.
@article{krzysztof2021applying,
title={Applying a Quantum Annealing Based Restricted Boltzmann Machine for MNIST Handwritten Digit Classification},
author={Krzysztof, Kurowski and Mateusz, Slysz and Marek, Subocz and Rafa{\l}, R{\'o}{\.z}ycki},
journal={CMST},
volume={27},
number={3},
pages={99--107},
year={2021},
publisher={PSNC, Poznan Supercomputing and Networking Center}
}