Anomaly Detection

Proprietary anomaly detection algorithms

OCI Anomaly Detection algorithms, backed by more than 150 patents, detect anomalies earlier with fewer false alarms. These algorithms work together to ensure higher sensitivity and better false alarm avoidance than other machine learning (ML) approaches, such as neural nets and support vector machines.

Blog: The fascinating (nuclear) history behind Oracle’s new anomaly detection service

Intelligent data preprocessing

OCI Anomaly Detection provides multiple data processing techniques that account for errors and imperfections in real-world input data, such as from low-resolution sensors. It automatically identifies and fixes data quality issues—resulting in fewer false alarms, better operations, and more accurate results.

Custom-trained models

APIs help developers upload raw data, train the anomaly detection model using their own business-specific data, and detect anomalies from the stored model. This makes creating highly accurate, custom-trained anomaly detection models accessible to everyone—even without data science experience.

Open-source options

Easy access to open-source technologies expands usage of OCI Anomaly Detection’s models. Pull time-series data from InfluxDB or streaming data from Apache Flink. Use open-source libraries like Plotly, Bokeh, and Altair for visualizations and to increase automation.

Ready-to-go results

OCI Anomaly Detection outputs include identified anomalies, ML model-based estimated values, and anomaly scores. Developers use these results to assess the severity of identified anomalies and automate business workflows to address them immediately.

Easy to integrate and deploy

OCI Anomaly Detection is a multitenant service over public REST APIs. Developers can deploy a scalable anomaly detection service easily without in-house data science and ML support, all with the lowest cost platform for data networking, storage, and egress.

Scalability on demand

OCI Anomaly Detection automatically scales for training and detection needs across all data sources and loads. Developers can now focus on creating applications and solutions to achieve their business goals, without worrying about the infrastructure.