Noise-Assisted Quantum Autoencoder

Quantum autoencoder is an efficient variational quantum algorithm for data compression. However, previous autoencoders fail to compress and recover high-rank mixed states. In this work, we discuss the fundamental properties limitations of standard model in more depth, provide information-theoretic solution its recovering fidelity. Based on understanding, present a noise-assisted go beyond limitations, our can achieve high fidelity general input Appropriate noise channels are used make mixedness output consistent, setup determined by measurement results trash system. Compared with original model, information fully algorithm. addition circuit design (noise-assisted) adiabatic that be implemented annealers. We verified validity methods through compressing thermal states transverse field Ising Werner For pure state ensemble compression, also introduce projected