Probability-Based Event Detection Using Bayesian Belief Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In complex or less well-constrained systems, determining the presence of events, such as anomalies in biological signals, is challenging due to the lack of a known model, making it difficult to automate detection and control actions based on probability.
Innovation Solution
A probability-based detection method using Bayesian networks and belief networks to combine evidence from different timescales, allowing for automated detection of events and controlled actions based on determined probabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a known deterministic model is used to detect events, then measurement precision is improved, but adaptability to complex or less well-constrained systems deteriorates
Solution Approach 1:
The patent transforms the deterministic model into a probabilistic model by changing the fundamental parameter from certainty to probability. This allows the system to handle complex or less well-constrained systems where events cannot be predicted with certainty, while maintaining rigorous mathematical foundations through Bayesian probability theory.
Solution Approach 2:
The patent introduces belief networks as an intermediary layer between the measured parameters and the event detection decision. This intermediary structure allows for the combination of multiple uncertain pieces of evidence in a systematic way, enabling the system to achieve reliable detection in complex systems where direct deterministic relationships do not exist.
2Productivity
If automation is increased to reduce human expert review, then productivity is improved, but reliability of detection deteriorates due to lack of human judgment
Solution Approach 1:
The patent implements feedback mechanisms where the belief network continuously updates probabilities based on new evidence and previous detections. This feedback loop allows the automated system to learn and adapt, maintaining high reliability while achieving full automation. The system can adjust its detection thresholds and priorities based on feedback from operational data.
Solution Approach 2:
The patent replaces the mechanical system of human expert review with an automated belief network system that uses probabilistic reasoning. This substitution maintains or improves reliability by applying consistent mathematical rules without human fatigue or bias, while dramatically increasing productivity through automated processing of multiple data streams.
3Measurement precision
If multiple parameters are measured to improve detection accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent segments the complex detection task into manageable components by organizing parameters into belief networks that represent different aspects or hypotheses. Each belief network processes specific subsets of parameters independently, then combines results through probabilistic reasoning. This segmentation reduces device complexity by modularizing the processing architecture.
Solution Approach 2:
The patent creates a universal belief network framework that can handle multiple types of parameters and detection scenarios through a single unified mathematical structure. This multi-functional approach allows the same core system to process diverse parameters (physiological signals, environmental data, operational metrics) without requiring separate specialized systems for each parameter type.
Data Source
Figure 1~2
Figure 3
Figure 4A~4D
AI summary
An apparatus comprising: means for determining a probability by combining at least: a probability that an event is present within a current feature of interest given a first set of previous features of interest, and a probability that the event is present within the current feature of interest given a second set of previous features of interest, different to the first set of previous features of interest; means for detecting the event based on the determined probability; and means for controlling, in dependence on the detection of the event, performance of an action.