# Just DIAL: DomaIn Alignment Layers for Unsupervised Domain Adaptation

## TL;DR

This paper introduces Domain Alignment Layers (DIAL), a novel method for unsupervised domain adaptation that aligns data distributions at the distribution level, improving visual recognition across different data settings without requiring target labels.

## Contribution

The paper proposes DIAL, a new approach that aligns source and target data distributions directly at the distribution level, differing from prior methods that focus on feature representation alignment.

## Key findings

- DIAL outperforms existing methods on three public benchmarks.
- DIAL effectively reduces domain shift without target labels.
- Experimental results confirm the robustness of DIAL across different datasets.

## Abstract

The empirical fact that classifiers, trained on given data collections, perform poorly when tested on data acquired in different settings is theoretically explained in domain adaptation through a shift among distributions of the source and target domains. Alleviating the domain shift problem, especially in the challenging setting where no labeled data are available for the target domain, is paramount for having visual recognition systems working in the wild. As the problem stems from a shift among distributions, intuitively one should try to align them. In the literature, this has resulted in a stream of works attempting to align the feature representations learned from the source and target domains. Here we take a different route. Rather than introducing regularization terms aiming to promote the alignment of the two representations, we act at the distribution level through the introduction of \emph{DomaIn Alignment Layers} (\DIAL), able to match the observed source and target data distributions to a reference one. Thorough experiments on three different public benchmarks we confirm the power of our approach.