FREEDOM: Target Label & Source Data & Domain Information-Free Multi-Source Domain Adaptation for Unsupervised Personalization

TL;DR

FREEDOM introduces a practical multi-source domain adaptation framework that operates without target labels, source data, or domain information, using generative models to achieve personalized models with reduced size.

Contribution

It proposes a novel Three-Free Domain Adaptation scenario and a framework that does not require domain labels or source data, advancing practical unsupervised personalization.

Findings

Achieves state-of-the-art performance without domain information.

Reduces final model size on the target side.

Operates independently of the number of source domains.

Abstract

From a service perspective, Multi-Source Domain Adaptation (MSDA) is a promising scenario to adapt a deployed model to a client's dataset. It can provide adaptation without a target label and support the case where a source dataset is constructed from multiple domains. However, it is impractical, wherein its training heavily relies on prior domain information of the multi-source dataset -- how many domains exist and the domain label of each data sample. Moreover, MSDA requires both source and target datasets simultaneously (physically), causing storage limitations on the client device or data privacy issues by transferring client data to a server. For a more practical scenario of model adaptation from a service provider's point of view, we relax these constraints and present a novel problem scenario of Three-Free Domain Adaptation, namely TFDA, where 1) target labels, 2) source dataset, and mostly 3) source domain information (domain labels + the number of domains) are unavailable. Under the problem scenario, we propose a practical adaptation framework called FREEDOM. It leverages the power of the generative model, disentangling data into class and style aspects, where the style is defined as the class-independent information from the source data and designed with a nonparametric Bayesian approach. In the adaptation stage, FREEDOM aims to match the source class distribution with the target's under the philosophy that class distribution is consistent even if the style is different; after then, only part of the classification model is deployed as a personalized network. As a result, FREEDOM achieves state-of-the-art or comparable performance even without domain information, with reduced final model size on the target side, independent of the number of source domains.