Minimal-Entropy Correlation Alignment for Unsupervised Deep Domain Adaptation

TL;DR

This paper introduces a novel deep learning method for unsupervised domain adaptation that aligns second order statistics along geodesics, improving performance on standard benchmarks.

Contribution

It proposes a new entropy minimization approach based on geodesic alignment of second order statistics, differing from Euclidean methods.

Findings

Outperforms existing methods on domain adaptation benchmarks.

Effectively aligns source and target domain distributions.

Demonstrates the importance of geodesic alignment in unsupervised adaptation.

Abstract

In this work, we face the problem of unsupervised domain adaptation with a novel deep learning approach which leverages on our finding that entropy minimization is induced by the optimal alignment of second order statistics between source and target domains. We formally demonstrate this hypothesis and, aiming at achieving an optimal alignment in practical cases, we adopt a more principled strategy which, differently from the current Euclidean approaches, deploys alignment along geodesics. Our pipeline can be implemented by adding to the standard classification loss (on the labeled source domain), a source-to-target regularizer that is weighted in an unsupervised and data-driven fashion. We provide extensive experiments to assess the superiority of our framework on standard domain and modality adaptation benchmarks.