Decorrelated Clustering with Data Selection Bias

TL;DR

This paper introduces DCKM, a clustering algorithm that mitigates selection bias effects by decorrelating features through sample weighting, leading to improved clustering performance on biased data.

Contribution

The paper proposes a novel decorrelation regularizer integrated with k-means to address selection bias in clustering, a largely unexplored area.

Findings

DCKM outperforms existing methods on real-world datasets.

Removing feature correlations improves clustering accuracy.

The approach effectively balances sample distribution.

Abstract

Most of existing clustering algorithms are proposed without considering the selection bias in data. In many real applications, however, one cannot guarantee the data is unbiased. Selection bias might bring the unexpected correlation between features and ignoring those unexpected correlations will hurt the performance of clustering algorithms. Therefore, how to remove those unexpected correlations induced by selection bias is extremely important yet largely unexplored for clustering. In this paper, we propose a novel Decorrelation regularized K-Means algorithm (DCKM) for clustering with data selection bias. Specifically, the decorrelation regularizer aims to learn the global sample weights which are capable of balancing the sample distribution, so as to remove unexpected correlations among features. Meanwhile, the learned weights are combined with k-means, which makes the reweighted k-means cluster on the inherent data distribution without unexpected correlation influence. Moreover, we derive the updating rules to effectively infer the parameters in DCKM. Extensive experiments results on real world datasets well demonstrate that our DCKM algorithm achieves significant performance gains, indicating the necessity of removing unexpected feature correlations induced by selection bias when clustering.