"Verifying Classification with Limited Disclosure" - Siddharth Bhandari, Research at TTIC
Автор: TTIC
Загружено: 2025-05-12
Просмотров: 63
“Verifying Classification with Limited Disclosure”
Siddharth Bhandari, Toyota Technological Institute at Chicago (TTIC)
Originally recorded on May 2, 2025, at TTIC, 6045 S. Kenwood Avenue, Chicago, IL.
In this talk, Siddharth Bhandari presents a theoretical framework for legal e-discovery as a multi-party classification and verification task. He introduces an interactive protocol that ensures perfect recall of responsive documents, minimizes non-relevant disclosure, and incentivizes truthfulness. The talk connects this framework to the Leave-One-Out dimension of hypothesis classes and introduces a novel geometric bound for linear classifiers.
Timestamps:
00:00 Introduction
03:10 Talk begins
#Theory #Algorithms #MachineLearning #Verification #Classification #PhDThesis #ComputerScience #LegalTech #TTIC #Research #Ediscovery #HypothesisClasses #VectorPacking
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
