EXoN: EXplainable encoder Network

TL;DR

EXoN introduces an explainable semi-supervised VAE that allows manual latent space design, improving classification and interpretability without sacrificing generative quality, demonstrated on MNIST and CIFAR-10 datasets.

Contribution

The paper presents EXoN, a novel semi-supervised VAE with an explainable and customizable latent space, integrating SCI for enhanced classification performance.

Findings

Produces an explainable latent space that aligns with labeled data.

Reduces analysis cost of latent representations.

Maintains generative quality while improving classification.

Abstract

We propose a new semi-supervised learning method of Variational AutoEncoder (VAE) which yields a customized and explainable latent space by EXplainable encoder Network (EXoN). Customization means a manual design of latent space layout for specific labeled data. To improve the performance of our VAE in a classification task without the loss of performance as a generative model, we employ a new semi-supervised classification method called SCI (Soft-label Consistency Interpolation). The classification loss and the Kullback-Leibler divergence play a crucial role in constructing explainable latent space. The variability of generated samples from our proposed model depends on a specific subspace, called activated latent subspace. Our numerical results with MNIST and CIFAR-10 datasets show that EXoN produces an explainable latent space and reduces the cost of investigating representation patterns on the latent space.