Latent space conditioning for improved classification and anomaly detection
This work addresses the challenge of unsupervised anomaly detection and clustering for data with labels, offering a domain-specific improvement over existing techniques.
The paper tackles the problem of improving clustering and anomaly detection by proposing a conditional latent space variational autoencoder that separates the latent space based on data information, fitting a Gaussian mixture model prior per class. The method is evaluated using V-score for clustering and a new unsupervised metric for anomaly detection, showing comparisons against established methods like isolation forest and one-class SVM.
We propose a new type of variational autoencoder to perform improved pre-processing for clustering and anomaly detection on data with a given label. Anomalies however are not known or labeled. We call our method conditional latent space variational autonencoder since it separates the latent space by conditioning on information within the data. The method fits one prior distribution to each class in the dataset, effectively expanding the prior distribution to include a Gaussian mixture model. Our approach is compared against the capabilities of a typical variational autoencoder by measuring their V-score during cluster formation with respect to the k-means and EM algorithms. For anomaly detection, we use a new metric composed of the mass-volume and excess-mass curves which can work in an unsupervised setting. We compare the results between established methods such as as isolation forest, local outlier factor and one-class support vector machine.