Method for realizing dynamic and secure synthetic data generation, involves converting data point from feature domain to frequency domain using Fast Fourier Transform, and computing total privacy budget total as combined effect of Fourierbased perturbation and Cholesky-preserving noise injection
	GANGRADE S; GANGRADE J; KUMAR S
	2025-03-27
专利权人	UNIV MANIPAL JAIPUR (UYMA-Non-standard)
申请日期	2025-03-27
专利号	IN202511028940-A
成果简介	NOVELTY - The method involves converting data point from feature domain to frequency domain using Fast Fourier Transform. Cholesky decomposition is performed. Lower triangular matrix that models feature relationships is obtained. Atructured noise is introduced. Synthetic data retains interfeature dependencies while preserving privacy is ensured. Total privacy budget total is computed as combined effect of Fourierbased perturbation and Cholesky-preserving noise injection, where synthetic dataset remains structurally faithful while preventing reverse engineering of sensitive information by integrating Fourier-domain perturbation with Cholesky covariance-preserving noise injection. USE - Method for realizing dynamic and secure synthetic data generation. ADVANTAGE - The method enables blending advanced mathematical techniques such as Fourier transforms and Cholesky decomposition to perturb data that preserves key structural properties while obfuscating sensitive information. The method enables providing a powerful shield against privacy breaches by integrating Fourier perturbation and maintaining covariance structure of original data, thus maintaining data utility while preserving privacy. DESCRIPTION OF DRAWING(S) - The drawing shows a flowchart illustrating a method for realizing dynamic and secure synthetic data generation.
IPC 分类号	G06F-017/16 ; G06F-021/62 ; G06N-007/01 ; H04L-009/40 ; H04W-012/02
国家	印度
专业领域	信息技术
语种	英语
成果类型	专利
文献类型	科技成果
条目标识符	http://119.78.100.226:8889/handle/3KE4DYBR/13483
专题	中国科学院新疆生态与地理研究所
作者单位	UNIV MANIPAL JAIPUR (UYMA-Non-standard)
推荐引用方式 GB/T 7714	GANGRADE S,GANGRADE J,KUMAR S. Method for realizing dynamic and secure synthetic data generation, involves converting data point from feature domain to frequency domain using Fast Fourier Transform, and computing total privacy budget total as combined effect of Fourierbased perturbation and Cholesky-preserving noise injection. IN202511028940-A[P]. 2025.

条目包含的文件
条目无相关文件。

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。