This research paper discusses the potential of Quantum Machine Learning (QML) and the challenges that come with it, such as vanishing gradients and efficient encoding methods. The authors propose a method for reversed training, aiming to increase single-sample accuracy. They demonstrate the effectiveness of their method by testing it on variational quantum circuits (VQCs), achieving an increase of 10-15% in single-sample inference accuracy. The study provides significant insights into the development of more efficient QML models in the future.
Publication date: 16 Oct 2023
Project Page: https://arxiv.org/abs/2310.10629v1
Paper: https://arxiv.org/pdf/2310.10629
