الفهرس 
Background and Related Work
A New Anomaly Detection Framework using Two-frame Motion Estimation Algorithm and Residual AutoencoderOnly 14 pages are availabe for public view
 |  | from | 26 |  |  |
| | | from | 26 | | |
Abstract
Anomaly detection plays a pivotal role that has been studied in several application domains such as system health monitoring, creditcard fraud detection, and computer network intrusion detection. It is noted that, the data quality management is a critical issue in the internet of things (IoT) applications to provide high quality data for decision making. Anomaly detection methods are employed to detect anomalies to improve the data quality by extracting patterns that do not match to the expected behavior in the data. The main challenge in detecting anomalies is the difficulty to differentiate between the normal and abnormal events, because the boundary is usually imprecise between the normal and abnormal behaviors. Another challenge is the difficulty to obtain labeled data which is necessary to train the system to detect anomalies.