2026-03-09T15:57:00Z https://ipsj.ixsq.nii.ac.jp/oai

oai:ipsj.ixsq.nii.ac.jp:00220425 2025-01-19T14:34:35Z 1164:10193:10905:11012

Quantum Generative Model with Optimal Transport Quantum Generative Model with Optimal Transport Hiroyuki, Tezuka Shumpei, Uno Naoki, Yamamoto Hiroyuki, Tezuka Shumpei, Uno Naoki, Yamamoto Generative model is an unsupervised machine learning framework, that exhibits strong performance in imaging or anomaly detection in classical machine learning regime. Recently we find several quantum version of generative model, some of which are even proven to have quantum advantage. However, those proposals have a strict limitation; that is, the quantum state to be learned (i.e., the quantum state that the model produces) is limited to a single quantum state, and thus those methods are not applicable to a set of quantum states. In this paper, we propose a quantum generative model that can learn a set of quantum state, in an unsupervised machine learning framework. The key idea is to introduce a loss function calculated based on optimal transport distance, i.e. Wasserstein distance. We then apply the proposed method to an anomaly detection task, that cannot be handled via existing methods. The proposed model paves the way for a wide application such as the health check of quantum devices and efficient initialization of quantum computation. Generative model is an unsupervised machine learning framework, that exhibits strong performance in imaging or anomaly detection in classical machine learning regime. Recently we find several quantum version of generative model, some of which are even proven to have quantum advantage. However, those proposals have a strict limitation; that is, the quantum state to be learned (i.e., the quantum state that the model produces) is limited to a single quantum state, and thus those methods are not applicable to a set of quantum states. In this paper, we propose a quantum generative model that can learn a set of quantum state, in an unsupervised machine learning framework. The key idea is to introduce a loss function calculated based on optimal transport distance, i.e. Wasserstein distance. We then apply the proposed method to an anomaly detection task, that cannot be handled via existing methods. The proposed model paves the way for a wide application such as the health check of quantum devices and efficient initialization of quantum computation. technical report 情報処理学会 2022-10-20 application/pdf 研究報告量子ソフトウェア（QS） 21 2022-QS-7 1 11 2435-6492 AA12894105 https://ipsj.ixsq.nii.ac.jp/record/220425/files/IPSJ-QS22007021.pdf eng