WE2.R2.4

Efficient Unbiased Sparsification

Leighton Barnes, Timothy Chow, Emma Cohen, Keith Frankston, Benjamin Howard, Fred Kochman, Daniel Scheinerman, Jeffrey VanderKam, Center for Communications Research - Princeton, United States
Session:
Semi-supervised and Federated Learning
Track:
8: Machine Learning
Location:
Ypsilon I-II-III
Presentation Time:
Wed, 10 Jul, 12:30 - 12:50
Session Chair:
Gholamali Aminian, Alan Turing Institute
Abstract
An unbiased $m$-sparsification of a vector $p\in \R^n$ is a random vector $Q\in \R^n$ with mean~$p$ that has at most $m
Session WE2.R2
WE2.R2.1: Fast FixMatch: Faster Semi-Supervised Learning with Curriculum Batch Size
John Chen, Chen Dun, Anastasios Kyrillidis, Rice University, United States
WE2.R2.2: Robust Semi-supervised Learning via f-Divergence and ɑ-Rényi Divergence
Gholamali Aminian, The Alan Turing Institute, United Kingdom; Amirhossien Bagheri, Sharif University Technology, Iran; Mahyar JafariNodeh, Massachusetts Institute of Technology, United States; Radmehr Karimian, Mohammad-Hossein Yassaee, Sharif University Technology, Iran
WE2.R2.3: Fed-IT: Addressing Class Imbalance in Federated Learning through an Information-Theoretic Lens
Shayan Mohajer Hamidi, Renhao Tan, Linfeng Ye, En-Hui Yang, University of Waterloo, Canada
WE2.R2.4: Efficient Unbiased Sparsification
Leighton Barnes, Timothy Chow, Emma Cohen, Keith Frankston, Benjamin Howard, Fred Kochman, Daniel Scheinerman, Jeffrey VanderKam, Center for Communications Research - Princeton, United States
Resources
View Manuscript