Anomaly Detector Configuration Using Model-Correlated Benchmarking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Anomaly detection in industrial contexts is challenging due to the scarcity of labeled data and the difficulty in identifying a combination of algorithms and parameters that yield useful results, leading to high costs and inefficiencies in production environments.

Innovation Solution

A methodological pipeline, called BRIGADE, is proposed to evaluate anomaly detection algorithms using a 2-fold cross-validation technique and a metric for ranked retrieval tasks, generating multiple benchmarking datasets from limited labeled positives and unlabelled production data, and selecting anomaly detectors based on correlation with a trained classifier's reference rankings.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If anomaly detection is performed across multiple data streams using traditional methods, then detection coverage is improved, but computational complexity and processing time increase exponentially

Engineering Contradiction:
Improveanomaly detection coverageVSAvoidcomputational complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments anomaly detection into stream-specific components by training individual anomaly detection models for each data stream type. Each model learns the normal behavior patterns of its specific stream, allowing parallel independent processing that reduces overall computational complexity while maintaining comprehensive detection coverage across all streams.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by focusing computational resources on detecting anomalies specific to each stream rather than applying a comprehensive but computationally expensive universal model to all streams. This selective approach processes only the necessary features for each stream type, reducing overall computational burden.

Inventive Principle:
Principle #16Partial or excessive action

2Speed

If stream-specific anomaly detection models are used, then processing speed is improved, but adaptability to new stream types deteriorates

Engineering Contradiction:
Improveanomaly processing speedVSAvoidadaptability to new stream types
Core Design Contradiction:
SpeedVSAdaptability or versatility

Solution Approach 1:

The patent creates a universal anomaly detection system that can handle multiple stream types through a common framework. The system uses a unified data structure and common processing pipeline that can accommodate different stream types, allowing the deployment of stream-specific models without sacrificing the ability to adapt to new stream types through the same framework.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Measurement precision

If comprehensive anomaly detection is implemented across all data streams, then detection accuracy is improved, but resource consumption increases

Engineering Contradiction:
Improveanomaly detection accuracyVSAvoidcomputational resource consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent segments the anomaly detection workload into independent stream-specific models that can be trained and executed separately. This segmentation allows the system to process each stream with appropriate computational resources rather than allocating excessive resources to all streams uniformly, reducing overall resource consumption while maintaining detection accuracy through specialized models.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4042654B1Dynamic configuration of anomaly detection
Publication Date: 2026.04.29 MICROSOFT TECHNOLOGY LICENSING LLC
  • EP4042654B1 patent drawingFigure 1
  • EP4042654B1 patent drawingFigure 2A
  • EP4042654B1 patent drawingFigure 2B

AI summary

The disclosed embodiments generate a plurality of anomaly detector configurations and compare results generated by these anomaly detectors to a reference result set. The reference result set is generated by a trained model. A correlation between each result generated by the anomaly detectors and the result set is compared to select an anomaly detector configuration that provides results most similar to those of the trained model. In some embodiments, data defining the selected configuration is then communicated to a product installation. The product installation instantiates the defined anomaly detector and analyzes local events using the instantiated detector. In some other embodiments, the defined anomaly detector is instantiated by the same system that selects the anomaly detector, and thus in these embodiments, the anomaly detector configuration is not transmitted from one system to another.