OAAE: Adversarial Autoencoders for Novelty Detection in Multi-modal Normality Case via Orthogonalized Latent Space
This work provides an incremental improvement for novelty detection in multi-modal normality cases, which is a problem for practitioners working with complex datasets.
The paper addresses novelty detection in multi-modal normality cases by proposing a new novelty score based on an orthogonalized latent space. It disentangles features in the latent space using mutual class information through orthogonal low-rank embedding, defining novelty by the change of each latent vector. The proposed algorithm outperforms state-of-the-art GAN-based novelty detection algorithms like RaPP and OCGAN.
Novelty detection using deep generative models such as autoencoder, generative adversarial networks mostly takes image reconstruction error as novelty score function. However, image data, high dimensional as it is, contains a lot of different features other than class information which makes models hard to detect novelty data. The problem gets harder in multi-modal normality case. To address this challenge, we propose a new way of measuring novelty score in multi-modal normality cases using orthogonalized latent space. Specifically, we employ orthogonal low-rank embedding in the latent space to disentangle the features in the latent space using mutual class information. With the orthogonalized latent space, novelty score is defined by the change of each latent vector. Proposed algorithm was compared to state-of-the-art novelty detection algorithms using GAN such as RaPP and OCGAN, and experimental results show that ours outperforms those algorithms.