Insight
Identifying Anomalous Behavior using the Anomaly Surfer
- 14 April 2025
- 5 min
Anomaly detection plays a critical role in various industries, where the presence of abnormal behavior or unexpected patterns can indicate potential problems or opportunities. For example, in the manufacturing industry, anomaly detection can help identify equipment malfunctions and prevent downtime. However, the accurate detection of anomalies is difficult for a number of reasons, such as the stochastic nature of sensor data, high data dimensionality, and the difficulty in defining what is normal and what is abnormal behavior.
Anomaly Detection
In this whitepaper, we introduce a technology developed to increase anomaly detection accuracy on time series data: the Anomaly Surfer, a light-weight deep learning anomaly detection model based on the wavelet transform and maximum likelihood estimation. We show that our model performs comparatively better than other state-ofthe-art models by benchmarking it on real-life and synthetic datasets often used in research. Furthermore, we identify multiple practical use cases in which the Anomaly Surfer technology could be of great value.
Whitepaper: Identifying Anomalous Behavior using the Anomaly Surfer
This whitepaper will describe:
- Introduction
- The problem and types of anomalies
- The Anomaly Surfer: background and our model
- Experiments and results
- Implications and use-cases
- Conclusion
