Tree-Based Ordinal Graphical Event Model for Temporal Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing graphical event models struggle to compactly capture ordinal historical dependence in temporal event data, leading to exponential complexity in managing potential orders of events, which hinders effective modeling and analysis in applications like system reliability, social networks, and finance.
Innovation Solution
A tree-based ordinal graphical event model is introduced, which learns to represent temporal relationships and ordinal historical dependence by constructing a tree structure that compactly captures conditional intensity parameters, leveraging a masking function to retain distinct label occurrences and induce a unique event label order, thereby simplifying the representation of event dynamics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional graphical event models are used to model temporal relationships, then the model can capture event dependencies, but the complexity of managing potential orders of events grows exponentially
Solution Approach 1:
The patent segments the exponential space of event orders by introducing a tree-based structure that hierarchically organizes event sequences. The tree decomposes complex temporal dependencies into manageable branches, where each node represents a subset of possible event orders. This segmentation allows the model to capture ordinal historical dependence without enumerating all possible permutations, thereby reducing model complexity while maintaining reliability in capturing temporal relationships.
2Loss of information
If all potential event orders are explicitly modeled, then complete temporal dependence is captured, but the quantity of parameters increases exponentially
Solution Approach 1:
The patent merges multiple event orders that share common historical dependencies into shared tree nodes. By combining parameter representations across equivalent event sequences, the model captures complete temporal dependence information while avoiding redundant parameters. The parameter sharing architecture allows different event orders to leverage common learned representations, significantly reducing the total parameter quantity while preserving full temporal dependence capture.
Solution Approach 2:
The tree-based OGEM structure serves multiple functions simultaneously: it organizes event sequences, shares parameters across equivalent histories, captures ordinal dependence, and enables efficient inference. This multi-functionality allows a single unified structure to replace what would otherwise require multiple separate parameter sets for different event orders, reducing overall parameter quantity while maintaining comprehensive temporal modeling capability.
Data Source
AI summary
A computer system, computer program product, and computer-implemented method are provided that includes learning a tree ordered graphical event model from an event dataset. Temporal relationships between one or more events in received temporal event data is modeled, and an ordered graphical event model (OGEM) graph is learned. The learned OGEM graph is configured to capture ordinal historical dependence. Leveraging the learned OGEM graph, a parameter sharing architecture is learned, including order dependent statistical and causal co-occurrence relationships among event types. A control signal to an operatively coupled event device that is associated with at least one event type reflected in the learned parameter sharing environment is dynamically issued. The control signal is configured to selectively control an event injection.