ScaleOut Software announces machine learning capabilities for its digital twin streaming services
William Bain of ScaleOut Software
ScaleOut Software announced extensions to its ScaleOut Digital Twin Streaming Service that enable real-time digital twin software to implement and host machine learning and statistical analysis algorithms that immediately identify unexpected behaviors exhibited by incoming telemetry. Real-time digital twins can now make extensive use of Microsoft’s ML.NET machine learning library to implement these groundbreaking capabilities for virtually any IoT device or source object.
Integration of machine learning with real-time digital twins offers powerful new options for real-time monitoring across a wide variety of applications. For example, cloud-based real-time digital twins can track a fleet of trucks to identify subtle changes in key engine parametres with predictive analytics that avoid costly failures. Security monitors tracking perimetre entrances and sound sensors can use machine learning techniques to automatically identify unexpected behaviors and generate alerts.
By harnessing the no-code ScaleOut Model Development Tool, a real-time digital twin can easily be enhanced to automatically analyse incoming telemetry messages using machine learning techniques. Machine learning provides important real-time insights that enhance situational awareness and enable fast, effective responses. The tool provides three configuration options for analysing numeric parametres contained within incoming messages to spot issues as they arise:
- Spike detection: Tracks a single parametre from a data source to identify a spike in its values over time using an adaptive kernel density estimation algorithm implemented by ML.NET.
- Trend detection: Also tracks a single parametre to identify a trend change, such as an unexpected increase over time for a parametre that is normally stable, using a linear regression algorithm that detects inflection points.
- Multi-variable anomaly detection: Tracks a set of related parametres in aggregate to identify anomalies using a user-selected machine-learning algorithm implemented by ML.NET that performs binary classification with supervised learning.
Once configured through the ScaleOut Model Development Tool, the ML algorithms run automatically and independently for each data source within their corresponding real-time digital twins as incoming messages are received. Each real-time digital twin can automatically capture anomalous events for follow-up analysis and generate alerts to popular alerting providers, such as Splunk, Slack, and Pager Duty, to support remediation by service or security teams.
“We are excited to offer powerful machine learning capabilities for real-time digital twins that will make it even easier to immediately spot issues or identify opportunities across a large population of data sources,” says, Dr. William Bain, ScaleOut Software’s CEO and founder. “ScaleOut Software has built the next step in the evolution of the Microsoft Azure IoT and ML.NET ecosystem, and we look forward to helping our customers harness these technologies to enhance their real-time monitoring and streaming analytics.”
Benefits of scaleOut’s real-time digital twins with machine learning
Integrating machine learning into ScaleOut’s real-time digital twins offers these key benefits:
- Powerful new capabilities for tracking data sources: The use of machine learning dramatically enhances the ability of streaming analytics running in real-time digital twins to automatically predict and identify emerging issues, thereby boosting their effectiveness.
- Simultaneous tracking for thousands of data sources: The integration of machine learning with real-time digital twins using in-memory computing techniques enables thousands of data streams to be independently analysed in real-time with fast, scalable performance.
- Fast, easy application deployment: With the ScaleOut Model Development Tool, these new machine learning capabilities can be configured in minutes using an intuitive GUI. No code development or library integration is required. Applications can optionally take advantage of a fully integrated rules engine to enhance their real-time analytics.
- Seamless use of Microsoft’s powerful machine learning library: Users can automatically take advantage of Microsoft’s technology for machine learning (ML.NET) to enhance their real-time device tracking and streaming analytics.
- Virtually unlimited application: These new capabilities are useful across a wide variety of applications that track numeric telemetry, with use cases including telematics, logistics, security, healthcare, retail, financial services, and many others.
For more information, please visit here.
Comment on this article below or via Twitter @IoTGN