Summary of Online Distribution Learning with Local Private Constraints, by Jin Sima and Changlong Wu and Olgica Milenkovic and Wojciech Szpankowski
Online Distribution Learning with Local Private Constraints
by Jin Sima, Changlong Wu, Olgica Milenkovic, Wojciech Szpankowski
First submitted to arxiv on: 1 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Cryptography and Security (cs.CR); Data Structures and Algorithms (cs.DS); Information Theory (cs.IT)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper tackles online conditional distribution estimation with unbounded label sets under local differential privacy. The goal is to estimate an unknown function in real-time, generating an estimate of that function given a context and privatized labels. The researchers aim to minimize the cumulative KL-risk over a finite horizon. Under (ε,0)-local differential privacy, the paper shows that the KL-risk grows as 🕺(1/ε√KT) up to poly-logarithmic factors, which is different from the bound demonstrated for bounded label sets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Online conditional distribution estimation with unbounded label sets is important. The goal is to estimate an unknown function based on a context and some information about that function. The researchers want to do this while keeping the information private. They show that under certain conditions, the risk of not estimating the function correctly grows as 1/ε√KT. |
