Sign In

Communications of the ACM

ACM TechNews

New Dependent Routing Technique Helps Eliminate Algorithmic Bias

View as: Print Mobile App Share:
Guaranteeing fairness in algorithmic resource allocation.

Researchers at the University of Maryland developed a Symmetric Randomized Dependent Rounding technique, to ensure outliers are included in cluster sorting.


Researchers at the University of Maryland have developed a new method to counter algorithmic bias, with the goal of proving algorithms can guarantee fairness in resource allocation.

The researchers applied a two-stage dependent rounding technique, which takes the form of "slowing down" and "early stopping" of algorithmic calculations.

They note the general problem in algorithmic sorting of variables is that outliers tend to get left out of cluster sorting. The team says the further from the center, the less likely that a specific data point will be included in the cluster that is analyzed, reinforcing a bias against statistical variants.

To solve this problem, the researchers created a new Symmetric Randomized Dependent Rounding technique, which modifies existing algorithms to update variables symmetrically, with additional randomization. The team say this ensures markers far from the center are included in the ongoing algorithmic correlation.

From The Stack (UK)
View Full Article


Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA


No entries found