Predicting bias in machine learned classifiers using clustering
Loading...
Authors
Thomson, Robert
Alhajjar, Elie
Irwin, Joshua
Russell, Travis
Issue Date
2018
Type
Conference presentations, papers, posters
Language
Keywords
bias , classification
Alternative Title
Abstract
We investigate the problem of diagnosing bias in machine learned classifiers by examining performance on the clusters of a testing set. We propose an algorithm for predicting and mitigating the relative bias in the classifier. We examine the performance of this algorithm in a few well-studied datasets and discuss potential applications.
Description
Citation
Thomson, Robert, Elie Alhajjar, Joshua Irwin, and Travis Russell. "Predicting bias in machine learned classifiers using clustering." In Annual social computing, behavior prediction, and modeling-behavioral representation in modeling simulation conference. 2018.
Publisher
SBP-BRiMS Annual Conference