Probabilistic Combination of Classifier and Cluster Ensembles for Non-transductive Learning

TL;DR

This paper introduces a Bayesian framework that combines classifier and cluster ensemble outputs to improve non-transductive learning, especially under concept drift and data privacy constraints.

Contribution

It proposes a novel probabilistic method that integrates classifier and cluster ensemble information for better target data classification in non-transductive settings.

Findings

Outperforms classifier ensemble methods in experiments

Effective in detecting distribution shifts

Supports privacy-aware and distributed learning

Abstract

Unsupervised models can provide supplementary soft constraints to help classify new target data under the assumption that similar objects in the target set are more likely to share the same class label. Such models can also help detect possible differences between training and target distributions, which is useful in applications where concept drift may take place. This paper describes a Bayesian framework that takes as input class labels from existing classifiers (designed based on labeled data from the source domain), as well as cluster labels from a cluster ensemble operating solely on the target data to be classified, and yields a consensus labeling of the target data. This framework is particularly useful when the statistics of the target data drift or change from those of the training data. We also show that the proposed framework is privacy-aware and allows performing distributed learning when data/models have sharing restrictions. Experiments show that our framework can yield superior results to those provided by applying classifier ensembles only.