Network Relationship Prediction via Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining relevant network relationships, such as graph algorithms and rule-based approaches, face challenges in accurately weighting edges or determining rules and thresholds for network features.
Innovation Solution
A computer-program product that trains a model to predict relevant network relationships by determining features and target variables for links between nodes, using a graph to label links as intra-community or inter-community, and applying the trained model to output relevant and non-relevant links.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If graph algorithm approach is used to determine network relationships, then network relationships can be identified, but determining relevant edge or link weights becomes challenging
Solution Approach 1:
The patent introduces an intermediary machine learning model that acts as a mediator between the graph algorithm approach and the network relationships. The model takes graph features as input and outputs predicted link weights, thereby resolving the complexity of directly determining edge weights while maintaining reliable network relationship identification
Solution Approach 2:
The patent replaces the mechanical/manual process of determining link weights with an automated machine learning system. The model automatically learns and predicts link weights based on training data, eliminating the need for manual weight assignment and reducing the complexity associated with graph algorithm approaches
2Reliability
If rule-based approach is used to determine network relationships, then relationships can be identified, but determining the rules and thresholds for network features becomes challenging
Solution Approach 1:
The machine learning model performs self-service by automatically learning the optimal rules and thresholds from training data during the model training phase. Instead of requiring manual rule creation, the model autonomously discovers patterns and relationships, thereby reducing the complexity of rules and thresholds determination while maintaining reliable network relationship identification
Solution Approach 2:
The patent transforms the static rule-based system into a dynamic parameter-driven system. The model learns optimal parameter values (thresholds and rules) from training data and adapts these parameters based on the specific characteristics of the network data, thereby simplifying the rule determination process while maintaining high reliability
3Ease of manufacture
If traditional methods are used for network relationship determination, then existing approaches can be applied, but accuracy in identifying relevant relationships deteriorates
Solution Approach 1:
The patent applies preliminary action by training the machine learning model in advance on labeled network data before deployment. This pre-training phase allows the model to learn accurate patterns and relationships, thereby improving measurement precision when the model is subsequently applied to identify relevant network relationships in operational settings
Data Source
AI summary
The computing device trains a first model on a first data set using a first graph to predict relevant links between a plurality of nodes. The computing device obtains the first data set or a second data set associated with the plurality of nodes. The computing device determines the one or more features for the one or more links between the plurality of nodes, applies the trained first model to the one or more links between the plurality of nodes, outputs the relevant links and non-relevant links of the one or more links between the plurality of nodes, removes the non-relevant links between the plurality of nodes, connects each node of the plurality of nodes with the relevant links to generate one or more first sets of networks, and outputs the one or more first sets of generated networks.