Quantum Physics
[Submitted on 3 Jul 2018 (v1), revised 22 Oct 2018 (this version, v3), latest version 16 Jun 2020 (v6)]
Title:Quantum generative adversarial network for generating discrete data
View PDFAbstract:Generative adversarial network (GAN) is an effective machine learning framework to train unsupervised generative models, and has drawn lots of attention in recent years. In the GAN framework, the generator is trained by an adversarial discriminator, in order to generate new samples that follows the probability distribution of a given training dataset. Classical GANs cannot generate discrete data due to the requirement of differentiability on the design of generators. Distributions of measurement outcomes of quantum circuits are continuous and differentiable, so quantum GANs can generate discrete data and complement classical GANs. In this paper, we present a quantum version of GAN for generation of discrete data, where parameterized quantum circuits are trained by a classical discriminator. Two families of quantum circuits, both composed of simple one-qubit rotation and two-qubit controlled-phase gates, are considered. The results of a small-scale proof-of-principle numerical experiment demonstrate that quantum circuits can be effectively trained in an adversarial way for generative tasks.
Submission history
From: Haozhen Situ [view email][v1] Tue, 3 Jul 2018 15:18:18 UTC (206 KB)
[v2] Tue, 17 Jul 2018 09:13:35 UTC (565 KB)
[v3] Mon, 22 Oct 2018 15:29:56 UTC (1,796 KB)
[v4] Tue, 11 Dec 2018 09:09:00 UTC (1,798 KB)
[v5] Tue, 22 Oct 2019 02:33:17 UTC (1,794 KB)
[v6] Tue, 16 Jun 2020 07:30:03 UTC (1,799 KB)
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.