Unsupervised Domain Adaptation with Copula Models

TL;DR

This paper introduces a copula-based regression framework for unsupervised domain adaptation, enabling more flexible modeling of conditional densities and effective feature mappings to reduce domain mismatch.

Contribution

It presents a novel copula-based approach leveraging Sklar's theorem to improve domain adaptation by modeling complex dependencies and transforming data for better transferability.

Findings

Outperforms recent feature transformation methods on benchmark datasets

Achieves more robust and accurate target label estimation

Effective in diverse regression models and applications like emotion estimation

Abstract

We study the task of unsupervised domain adaptation, where no labeled data from the target domain is provided during training time. To deal with the potential discrepancy between the source and target distributions, both in features and labels, we exploit a copula-based regression framework. The benefits of this approach are two-fold: (a) it allows us to model a broader range of conditional predictive densities beyond the common exponential family, (b) we show how to leverage Sklar's theorem, the essence of the copula formulation relating the joint density to the copula dependency functions, to find effective feature mappings that mitigate the domain mismatch. By transforming the data to a copula domain, we show on a number of benchmark datasets (including human emotion estimation), and using different regression models for prediction, that we can achieve a more robust and accurate estimation of target labels, compared to recently proposed feature transformation (adaptation) methods.