General Domain Adaptation Through Proportional Progressive Pseudo Labeling

TL;DR

This paper introduces PPPL, a simple and effective domain adaptation method that progressively improves target domain classification across various input types by selectively pseudo-labeling samples.

Contribution

The paper presents PPPL, a versatile domain adaptation technique applicable to multiple data types, with a straightforward implementation and improved generalization performance.

Findings

PPPL outperforms baseline methods on 6 diverse datasets.

PPPL effectively reduces classification error during training.

PPPL generalizes well across image, text, and anomaly detection tasks.

Abstract

Domain adaptation helps transfer the knowledge gained from a labeled source domain to an unlabeled target domain. During the past few years, different domain adaptation techniques have been published. One common flaw of these approaches is that while they might work well on one input type, such as images, their performance drops when applied to others, such as text or time-series. In this paper, we introduce Proportional Progressive Pseudo Labeling (PPPL), a simple, yet effective technique that can be implemented in a few lines of code to build a more general domain adaptation technique that can be applied on several different input types. At the beginning of the training phase, PPPL progressively reduces target domain classification error, by training the model directly with pseudo-labeled target domain samples, while excluding samples with more likely wrong pseudo-labels from the training set and also postponing training on such samples. Experiments on 6 different datasets that include tasks such as anomaly detection, text sentiment analysis and image classification demonstrate that PPPL can beat other baselines and generalize better.