SGN: A Similarity-based Generative Network for Data Generation under Distribution Shift
2026-07-20 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors propose a new method called Similarity-based Generative Network (SGN) that helps generate new data for a target domain using information learned from a source domain without needing to retrain the model. SGN creates a special space where data points are organized based on their labels and similarities, which helps keep generated samples consistent with the target domain's characteristics. By combining a small labeled set from the target domain in this space, the model can produce realistic and useful new examples without extra optimization. Their experiments show SGN works well on both images and tabular data when the source and target domains differ.
generative modelsdomain shiftdata augmentationlatent spaceencoder-decodersimilarity learningsource domaintarget domaintransfer learningdistribution shift
Authors
Jiaqi Zhu, Xincheng Chen, Yuncheng Wu, Zhaojing Luo, Beng Chin Ooi
Abstract
Generative models trained on a source domain often produce samples that are poorly aligned with shifted target domains, limiting their effectiveness for target-domain data augmentation. Although target-specific adaptation can reduce this mismatch, it typically requires additional optimization and domain-specific parameters. We propose a Similarity-based Generative Network (SGN), a reusable framework that is trained once on labeled source data and applied to new target domains without parameter updates. SGN learns a latent space structured by label-induced pairwise similarities while preserving reconstructive information through an encoder-decoder architecture. At generation time, a small labeled representative set from the target domain is encoded and combined in the learned latent space, allowing the generated samples to inherit target-specific characteristics while maintaining class consistency. We further analyze the realizability and dimensionality requirements of the proposed similarity structure. Experiments on image and tabular datasets demonstrate the effectiveness of SGN for target-guided data augmentation under source-to-target distribution shifts.