Anomaly Detection by Deep Learning
|
The purpose of anomaly detection using data is to show high performance during the learning process and Deployment.
However, it cannot be guaranteed the data used in the learning process have the same distribution as the data from sensors in the field.
The reason is that sensor data is affected by the temperature, humidity, conditions, etc., which vary greatly in the various environments.
The goal of our laboratory is to perform robust anomaly detection by relearning while maintaining the core part of the model so that if there is an anomaly detection model already built in one facility, it can perform well in anomaly detection even when it goes to another environment, and we are conducting research related to this.
|
