Semi-Supervised Noise Adaptation: Transferring Knowledge from Noise Domain
For semi-supervised learning practitioners, this work presents a surprising and potentially impactful approach by using noise as a source domain, though the practical benefits over existing methods remain to be seen.
This paper introduces Semi-Supervised Noise Adaptation (SSNA), a novel problem where a synthetic noise domain is used to improve target domain generalization in semi-supervised learning. The proposed Noise Adaptation Framework (NAF) tightens the generalization bound and achieves improved performance, with extensive experimental validation.
Transfer learning aims to facilitate the learning of a target domain by transferring knowledge from a source domain. The source domain typically contains semantically meaningful samples (*e.g.*, images) to facilitate effective knowledge transfer. However, a recent study observes that the noise domain constructed from simple distributions (*e.g.*, Gaussian distributions) can serve as a surrogate source domain in the semi-supervised setting, where only a small proportion of target samples are labeled while most remain unlabeled. Based on this surprising observation, we formulate a novel problem termed *Semi-Supervised Noise Adaptation* (SSNA), which aims to leverage a synthetic noise domain to improve the generalization of the target domain. To address this problem, we first establish a generalization bound characterizing the effect of the noise domain on generalization, based on which we propose a Noise Adaptation Framework (NAF). Extensive experiments demonstrate that NAF effectively leverages the noise domain to tighten the generalization bound of the target domain, leading to improved performance. The codes are available at https://github.com/AIResearch-Group/SSNA.