Greenhouse: A Zero-Positive Machine Learning System for Time-Series Anomaly Detection

TL;DR

Greenhouse is a novel machine learning system designed for time-series anomaly detection that operates without requiring positive examples during training.

Contribution

The paper introduces Greenhouse, a zero-positive learning approach for time-series anomaly detection, which is a new paradigm in this domain.

Findings

Preliminary results show effective anomaly detection without positive samples.

Greenhouse outperforms traditional methods requiring labeled anomalies.

The system demonstrates robustness across different time-series datasets.

Abstract

This short paper describes our ongoing research on Greenhouse - a zero-positive machine learning system for time-series anomaly detection.