Skip to content

Latent Pseudo-Anomaly Generation

Home / Concepts / Latent Pseudo-Anomaly Generation

Latent Pseudo-Anomaly Generation

Auto-generated stub. Edit this file to add more details.

A technique for unsupervised time-series anomaly detection that constructs pseudo-anomalies within the latent space, decoupling training from domain-specific hand-crafted rules.

Why It Matters

It addresses the bottleneck of requiring handcrafted anomalies in unsupervised time-series anomaly detection by shifting the problem to the latent space, which is a highly reusable methodology for representation learning.

Evidence

ASTER, a framework that generates pseudo-anomalies directly in the latent space, avoiding handcrafted anomaly injections and the need for domain expertise.

Metadata & Links