A Caltech Library Service

Robust Correction of Sampling Bias Using Cumulative Distribution Functions

Mazaheri, Bijan and Jain, Siddharth and Bruck, Jehoshua (2020) Robust Correction of Sampling Bias Using Cumulative Distribution Functions. Parallel and Distributed Systems Group Technical Reports, etr149. California Institute of Technology , Pasadena, CA. (Unpublished)

[img] PDF - Submitted Version
See Usage Policy.


Use this Persistent URL to link to this item:


Varying domains and biased datasets can lead to differences between the training and the target distributions, known as covariate shift. Current approaches for alleviating this often rely on estimating the ratio of training and target probability density functions. These techniques require parameter tuning and can be unstable across different datasets. We present a new method for handling covariate shift using the empirical cumulative distribution function estimates of the target distribution by a rigorous generalization of a recent idea proposed by Vapnik and Izmailov. Further, we show experimentally that our method is more robust in its predictions, is not reliant on parameter tuning and shows similar classification performance compared to the current state-of-the-art techniques on synthetic and real datasets.

Item Type:Report or Paper (Technical Report)
Related URLs:
URLURL TypeDescription Report
Jain, Siddharth0000-0002-9164-6119
Bruck, Jehoshua0000-0001-8474-0812
Additional Information:This work is supported by supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE-1745301, NSF Grant No. CCF-1717884 and The Carver Mead New Adventure Fund.
Group:Parallel and Distributed Systems Group
Funding AgencyGrant Number
NSF Graduate Research FellowshipDGE-1745301
Carver Mead New Adventures FundUNSPECIFIED
Series Name:Parallel and Distributed Systems Group Technical Reports
Issue or Number:etr149
Record Number:CaltechAUTHORS:20210624-211933517
Persistent URL:
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:109569
Deposited By: George Porter
Deposited On:24 Jun 2021 21:50
Last Modified:24 Jun 2021 21:50

Repository Staff Only: item control page