Mining Multi-Label Samples from Single Positive Labels

Youngin Cho, Daejin Kim, Jaegul Choo

Keywords: GANs, MCMC, sampling, conditional generation, single positive label, multi-label

Abstract: Conditional generative adversarial networks (cGANs) have shown superior results in class-conditional generation tasks. In order to simultaneously control multiple conditions, cGANs requires multi-label training datasets, where multiple labels can be assigned to each data instance. Nevertheless, the tremendous annotation cost limits the accessibility of multi-label datasets in the real world. Hence, we explore the practical setting called single positive setting, where each data instance is annotated by only one positive label with no explicit negative labels. To generate multi-label data in the single positive setting, we propose a novel sampling approach called single-to-multi-label (S2M) sampling, based on the Markov chain Monte Carlo method. As a widely applicable “add-on” method, our proposed S2M sampling enables existing unconditional and conditional GANs to draw high-quality multi-label data with a minimal annotation cost. Extensive experiments on real image datasets (e.g., CIFAR-10 and CelebA) verify the effectiveness and correctness of our method, even when compared to a model trained with fully annotated datasets.

One-sentence Summary: Our proposed S2M sampling can generate multi-label data only using single positive labels.

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/arxiv:2206.05764/code)