English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Very recently, studies have shown that quantum neural networks surpassclassical neural networks in tasks like image classification when a similarnumber of learnable parameters are used. However, the development andoptimization of quantum models are currently hindered by issues such as qubitinstability and limited qubit availability, leading to error-prone systems withweak performance. In contrast, classical models can exhibit high-performanceowing to substantial resource availability. As a result, more studies have beenfocusing on hybrid classical-quantum integration. A line of researchparticularly focuses on transfer learning through classical-quantum integrationor quantum-quantum approaches. Unlike previous studies, this paper introduces anew method to transfer knowledge from classical to quantum neural networksusing knowledge distillation, effectively bridging the gap between classicalmachine learning and emergent quantum computing techniques. We adapt classicalconvolutional neural network (CNN) architectures like LeNet and AlexNet toserve as teacher networks, facilitating the training of student quantum modelsby sending supervisory signals during backpropagation through KL-divergence.The approach yields significant performance improvements for the quantum modelsby solely depending on classical CNNs, with quantum models achieving an averageaccuracy improvement of 0.80% on the MNIST dataset and 5.40% on the morecomplex Fashion MNIST dataset. Applying this technique eliminates thecumbersome training of huge quantum models for transfer learning inresource-constrained settings and enables re-using existing pre-trainedclassical models to improve performance.Thus, this study paves the way forfuture research in quantum machine learning (QML) by positioning knowledgedistillation as a core technique for advancing QML applications.

Bridging Classical and Quantum Machine Learning: Knowledge Transfer From Classical to Quantum Neural Networks Using Knowledge Distillation