| PhD Seminar

Name of the Speaker: Mr. Gokularam M (EE17D400)
Guide: Prof. Giridhar K
Co-Guide: Prof. Sheetal Kalyani
Venue: ESB-244 (Seminar Hall)
Online meeting link: meet.google.com/nzo-uyzr-egc
Date/Time: 16th April 2024 (Tuesday), 3:00 PM
Title: A New Hybrid Noise Mechanism for Differential Privacy with Improved Utility.

Abstract

In the era of big data, the paramount concern is no longer the scarcity of information but the protection of our personal details. The framework of differential privacy renders analysis on the congregated data while ensuring that any individual's contribution to the dataset remains indistinguishable. Differential privacy is typically ensured by perturbation with additive noise that is sampled from a known distribution. In this talk, we will discuss an additive noise mechanism for differential privacy that adds noise sampled from a new noise density called flipped Huber distribution. It is a hybrid density that resembles Laplace in the centre and Gaussian in the tail. With a sharper centre and light, sub-Gaussian tail, this density has the best characteristics of both distributions and renders a better trade-off between privacy and accuracy than other existing mechanisms. We theoretically analyze the proposed mechanism, and we derive the necessary and sufficient condition in one dimension and a sufficient condition in high dimensions for the mechanism to guarantee (ϵ,δ)-differential privacy. Numerical simulations corroborate the efficacy of the proposed mechanism compared to other existing mechanisms in achieving a better trade-off between privacy and accuracy.