Random Quantum Neural Networks (RQNN) for Noisy Image Recog- nition

Author Konar D.; Gelenbe E.; Bhandary S.; Sarma A.; Cangi A.
Title Random Quantum Neural Networks (RQNN) for Noisy Image Recog- nition
Journal IEEE Xplore
Year 2022
Status Published
DOI 10.1109/QCE57702.2023.10240
Abstract <p>Classical Random Neural Networks (RNNs) have demonstrated effective applications in decision<br />
making, signal processing, and image recognition tasks. However, their implementation has been<br />
limited to deterministic digital systems that output probability distributions in lieu of stochastic be-<br />
haviors of random spiking signals. We introduce the novel class of supervised Random Quantum<br />
Neural Networks (RQNNs) with a robust training strategy to better exploit the random nature of<br />
the spiking RNN. The proposed RQNN employs hybrid classical-quantum algorithms with super-<br />
position state and amplitude encoding features, inspired by quantum information theory and the<br />
brain’s spatial-temporal stochastic spiking property of neuron information encoding. We have ex-<br />
tensively validated our proposed RQNN model, relying on hybrid classical-quantum algorithms via<br />
the PennyLane Quantum simulator with a limited number of qubits. Experiments on the MNIST,<br />
FashionMNIST, and KMNIST datasets demonstrate that the proposed RQNN model achieves an av-<br />
erage classification accuracy of 94.9%. Additionally, the experimental findings illustrate the proposed<br />
RQNN’s effectiveness and resilience in noisy settings, with enhanced image classification accuracy<br />
when compared to the classical counterparts (RNNs), classical Spiking Neural Networks (SNNs), and<br />
the classical convolutional neural network (AlexNet). Furthermore, the RQNN can deal with noise,<br />
which is useful for various applications, including computer vision in NISQ devices. The PyTorch<br />
code 1 is made available on GitHub to reproduce the results reported in this manuscript.</p>