Burn After Reading: Online Adaptation for Cross-domain Streaming Data

TL;DR

This paper introduces an online privacy-preserving framework that deletes data after processing and employs a novel cross-domain bootstrapping algorithm to improve unsupervised domain adaptation in streaming data scenarios.

Contribution

It proposes a new 'burn after reading' online framework combined with CroDoBo, a cross-domain bootstrapping method, to enhance privacy and adaptation without diverse source-target pairs.

Findings

Achieves state-of-the-art online performance on four benchmarks.

Effectively handles distribution shift with no sensitive data retention.

Introduces a multi-learner co-supervision training strategy.

Abstract

In the context of online privacy, many methods propose complex privacy and security preserving measures to protect sensitive data. In this paper, we argue that: not storing any sensitive data is the best form of security. Thus we propose an online framework that "burns after reading", i.e. each online sample is immediately deleted after it is processed. Meanwhile, we tackle the inevitable distribution shift between the labeled public data and unlabeled private data as a problem of unsupervised domain adaptation. Specifically, we propose a novel algorithm that aims at the most fundamental challenge of the online adaptation setting--the lack of diverse source-target data pairs. Therefore, we design a Cross-Domain Bootstrapping approach, called CroDoBo, to increase the combined diversity across domains. Further, to fully exploit the valuable discrepancies among the diverse combinations, we employ the training strategy of multiple learners with co-supervision. CroDoBo achieves state-of-the-art online performance on four domain adaptation benchmarks.