QRBM

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

Demo

To run demo, click on the badge below:

Dataset

To download the default dataset, click here: MNIST - a dataset of pictures of handwritten digits with 28x28 pixel resolution

Train the RBM yourself

The best way to run this code is to download the dataset and run all cells in the jupyter notebook.

Some results

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.

How to cite

@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}
}