Robust Semi-Supervised Classification using GANs with Self-Organizing Maps
This work addresses a specific limitation in semi-supervised GANs for classification, particularly in handling outliers, which is incremental as it builds on existing methods by integrating SOMs.
The paper tackles the problem of discriminating outliers from inliers (DOIC) in semi-supervised GANs for classification, proposing an architecture that combines self-organizing maps (SOMs) with GANs. Experimental results on hyperspectral image datasets show that incorporating SOMs substantially mitigates the DOIC problem compared to GANs without SOMs, with SS-GANs performing much better than supervised GANs on this issue.
Generative adversarial networks (GANs) have shown tremendous promise in learning to generate data and effective at aiding semi-supervised classification. However, to this point, semi-supervised GAN methods make the assumption that the unlabeled data set contains only samples of the joint distribution of the classes of interest, referred to as inliers. Consequently, when presented with a sample from other distributions, referred to as outliers, GANs perform poorly at determining that it is not qualified to make a decision on the sample. The problem of discriminating outliers from inliers while maintaining classification accuracy is referred to here as the DOIC problem. In this work, we describe an architecture that combines self-organizing maps (SOMs) with SS-GANS with the goal of mitigating the DOIC problem and experimental results indicating that the architecture achieves the goal. Multiple experiments were conducted on hyperspectral image data sets. The SS-GANS performed slightly better than supervised GANS on classification problems with and without the SOM. Incorporating the SOMs into the SS-GANs and the supervised GANS led to substantially mitigation of the DOIC problem when compared to SS-GANS and GANs without the SOMs. Furthermore, the SS-GANS performed much better than GANS on the DOIC problem, even without the SOMs.