A Univariate Time-series Anomaly Detection Method for Sensor Network based on Conditional Variational AutoEncoder

Key Points

• The research aims to develop a method for detecting anomalies in univariate time-series data from sensor networks.
• Developed a univariate time-series anomaly detection technique.
• Utilized conditional variational autoencoders for analysis.
• Tested the method on sensor network data.
• Demonstrated effectiveness in detecting anomalies in time-series data.
• Achieved higher accuracy compared to traditional methods.
• Showed improved reliability in monitoring sensor networks.

Abstract