Local differential privacy for tensors in distributed computing systems
Yachao Yuan, Xiao Tang, Yu Huang, Yingwen Wu, Jin Wang · Neurocomputing · 2026
Tensor-valued data, increasingly common in distributed big data applications like autonomous driving and smart healthcare, poses unique challenges for privacy protection due to its multidimensional structure and the risk of losing critical structural information. Traditional local differential privacy methods, designed for scalars and matrices, are insufficient for tensors, as they fail to preserve essential relationships among tensor elements. We propose TLDP, a novel LDP algorithm for Tensors that randomly decides whether to inject noise into each component with a certain probability. Such randomness substitutes for part of the perturbation noise and safeguards the raw data more effectively under the condition of guaranteeing differential privacy. To strike a better balance between utility and privacy, we incorporate a weight matrix that selectively protects sensitive regions. Both theoretical analysis and empirical findings from real-world datasets show that TLDP achieves superior utility while preserving privacy, making it a robust solution for high-dimensional tensor data. Our code is available at: https://github.com/fatmo666/TLDP.