{"componentChunkName":"component---src-templates-blog-post-js","path":"/blog/quantum-machine-learning-with-differential-privacy/","result":{"data":{"site":{"siteMetadata":{"title":"No Frills News"}},"contentfulNfnPost":{"postTitle":"Quantum machine learning with differential privacy","slug":"quantum-machine-learning-with-differential-privacy","createdLocal":"2023-02-11 14:30:46.845931","publishDate":"None","feedName":"Image Recognition","sourceUrl":{"sourceUrl":"https://www.nature.com/articles/s41598-022-24082-z"},"postSummary":{"childMarkdownRemark":{"html":"<p>Quantum computing basicsBecause of the power of superposition and entanglement generated by quantum gates, quantum computing can create a huge speedup in certain difficult computational tasks and afford quantum advantages to ML33,34.\n\\end{aligned} \\end{aligned}$$ (6)The general single-qubit rotation can be constructed with two of the single-qubit rotations (R<em>{x}), (R</em>{y}), and (R_{z}).\nFurthermore, differential privacy gives the worst-case scenario privacy loss, thus a smaller (\\varepsilon) does not necessarily mean the privacy is better.\nWe leave the development of a tighter upper bound on privacy loss in trained quantum circuits to a future study.\ndiscuss how adding depolarizable noise to a quantum circuit imposes differential privacy on the model, providing robustness against adversarial examples76.</p>"}}}},"pageContext":{"slug":"quantum-machine-learning-with-differential-privacy"}}}