Anomaly Detection in Geographic Areas via Dynamic Relationship Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional surveillance systems for monitoring geographic locations are limited in detecting anomalies as they require predefined, static data on normal relationships, which are impervious to changes and need continuous user input, making them less effective in dynamic environments.
Innovation Solution
A system and method that utilize an electronic computing device and monitoring devices to generate and compare expected relationship data with observed relationship data, using machine-learned models and object classifiers to identify anomalies by tracking entities and their trajectories within a geographic area, providing notifications to safety officers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If rule-based surveillance systems with predefined normal relationships are used, then anomaly detection capability is provided, but the system is limited in scope and cannot adapt to changes in the geographic location
Solution Approach 1:
The system transitions from static, predefined normal relationships to dynamic relationships that are continuously learned and updated through machine learning models. The normal relationship data is generated autonomously by analyzing historical surveillance data, allowing the system to adapt to changes in the geographic location without manual reconfiguration.
Solution Approach 2:
The system autonomously generates normal relationship data through machine learning models that analyze historical surveillance data. This eliminates the need for continuous manual input from users to update the definition of normal relationships, as the system self-updates its understanding of what constitutes normal activity patterns in the monitored location.
2Reliability
If rule-based systems with hardcoded normal relationships are used, then initial anomaly detection is enabled, but the system requires continued user input to accurately detect anomalies
Solution Approach 1:
The system autonomously generates and updates normal relationship data through machine learning models that continuously analyze historical surveillance data. This eliminates the need for continuous manual input from users to maintain accurate anomaly detection, as the system self-updates its understanding of normal activity patterns.
Solution Approach 2:
The system uses historical surveillance data as feedback to continuously refine and update the normal relationship data through machine learning. This feedback loop allows the system to improve its anomaly detection accuracy over time without requiring continuous manual intervention, as the model learns from past observations and adjusts its understanding of normal patterns.
3Ease of manufacture
If predefined normal relationship data is used, then anomaly detection can be implemented, but the relationships are impervious to changes within the geographic location
Solution Approach 1:
The system replaces static, hardcoded normal relationships with dynamic relationships that are continuously learned from historical surveillance data through machine learning models. This allows the system to adapt to changes in the geographic location while maintaining ease of implementation, as the adaptation occurs automatically through data-driven learning rather than manual reconfiguration.
Solution Approach 2:
The system changes the fundamental parameter of normal relationship data from fixed, predefined values to dynamically updated values generated by machine learning models. This parameter change enables the system to respond to location dynamics while maintaining simple deployment, as the models automatically adjust to new patterns without requiring manual parameter updates.
Data Source
AI summary
Methods for detecting anomalies in a geographic area include receiving, from an electronic computing device, expected relationship data indicating expected relationships between a plurality of entities within the geographic area; detecting the plurality of entities within the geographic area; generating observed relationship data indicating observed relationships between the plurality of entities; identifying the expected relationships between the plurality of entities based on the expected relationship data; determining that a given observed relationship between two entities of the plurality of entities is likely to represent an anomaly based on the expected relationship data; and providing an electronic notification to a safety officer, the electronic notification indicating that the given observed relationship is likely to represent the anomaly.