A Prototype-Oriented Framework for Unsupervised Domain Adaptation

TL;DR

This paper introduces a prototype-oriented framework for unsupervised domain adaptation that uses class prototypes for feature alignment, offering a memory-efficient, privacy-preserving alternative to statistical distance methods.

Contribution

It proposes a novel probabilistic framework that aligns target features with class prototypes without additional model parameters, applicable across various domain adaptation scenarios.

Findings

Achieves competitive performance with state-of-the-art methods.

Requires no extra model parameters and moderate computation increase.

Effective across multiple domain adaptation scenarios.

Abstract

Existing methods for unsupervised domain adaptation often rely on minimizing some statistical distance between the source and target samples in the latent space. To avoid the sampling variability, class imbalance, and data-privacy concerns that often plague these methods, we instead provide a memory and computation-efficient probabilistic framework to extract class prototypes and align the target features with them. We demonstrate the general applicability of our method on a wide range of scenarios, including single-source, multi-source, class-imbalance, and source-private domain adaptation. Requiring no additional model parameters and having a moderate increase in computation over the source model alone, the proposed method achieves competitive performance with state-of-the-art methods.