Research

Validating and testing a machine learning model is a critical stage in model development. For time-series anomaly detection, validation and testing is challenging because of the lack of labeled data and the difficulty of generating a realistic time-series with anomalies. Motivated by the continued success of Variational Auto-Encoders (VAE) and Generative Adversarial Networks (GANs) to produce realistic-looking data we provide a platform to generate a realistic time-series with anomalies called AnoGen. Our contribution includes a sampling technique that allows us to sample from the latent z space of a trained variational auto-encoder to deterministically generate a realistic time-series with anomalies.