ePrints@IIScePrints@IISc Home | About | Browse | Latest Additions | Advanced Search | Contact | Help

Interactive Inference under Information Constraints

Acharya, J and Canonne, CL and Liu, Y and Sun, Z and Tyagi, H (2021) Interactive Inference under Information Constraints. In: 2021 IEEE International Symposium on Information Theory, ISIT 2021, 12 - 20 July 2021, Virtual, Melbourne, pp. 326-331.

IEEE_ISIT 2021_326-331_2021.pdf - Published Version

Download (518kB) | Preview
Official URL: https://doi.org/10.1109/ISIT45174.2021.9518069


We consider the problem of distributed estimation and testing of discrete distributions under local information constraints that include communication and privacy as special cases. Our main result is a unified method that establishes tight bounds for interactive protocols under both the constraints and both the problems. Our main technical contribution is an average information bound which connects learning and testing and handles correlations due to interactivity. While we establish that for learning and testing under both the constraints above, interactivity does not help, we also illustrate a natural family of 'leaky query' local constraints under which interactive protocols strictly outperform the noninteractive ones for identity testing.

Item Type: Conference Paper
Publication: IEEE International Symposium on Information Theory - Proceedings
Publisher: Institute of Electrical and Electronics Engineers Inc.
Additional Information: The Copyright for this article belongs to the Authors.
Keywords: Average information; Discrete distribution; Distributed estimation; Identity testing; Interactive protocols; Local constraints; Local information; Technical contribution; Information theory
Department/Centre: Division of Electrical Sciences > Electrical Communication Engineering
Date Deposited: 14 Jul 2022 09:10
Last Modified: 14 Jul 2022 09:10
URI: https://eprints.iisc.ac.in/id/eprint/74400

Actions (login required)

View Item View Item