QQQYang/QNN-Baseline

provide benchmarks for multiple QNNs

The dilemma of quantum neural networks

This repository is the official implementation of The dilemma of quantum neural networks.

We systematically explore the learnability of quantum neural networks compared with classical neural networks. The overall hypothesis of quantum models and classical models is shown below:

Requirements

To install requirements:

pip install -r requirements.txt

Training & evaluation

Quantum neural networks

Running on a native linux server, please refer to script/run_example.sh:

python --config-file config/xxx.yml --index 0

To run experiments with different models and settings, just update the configuration file *.yml.

MLP

Please refer to classical.py

python classical.py

CNN

Run the following script:

python Simple_QCNN_MNIST.py

Datasets

• Quantum synthetic datasets: please refer to python file data_gen.py

python data_gen.py

• The Wine data: https://archive.ics.uci.edu/ml/datasets/wine
• MNIST: http://yann.lecun.com/exdb/mnist/

Bibtex

@article{qian2022dilemma,
  title={The dilemma of quantum neural networks},
  author={Qian, Yang and Wang, Xinbiao and Du, Yuxuan and Wu, Xingyao and Tao, Dacheng},
  journal={IEEE Transactions on Neural Networks and Learning Systems},
  year={2022},
  publisher={IEEE}
}