Hybrid Graph Prediction Framework for Complex Feature Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing graph convolutional neural network machine learning models struggle to accurately process complex graph data objects, failing to capture feature data related to node attributes, edge attributes, and edge weights, and are unable to holistically process interrelations of graph-based feature sources.
Innovation Solution
A hybrid graph-based processing machine learning framework that combines graph convolutional neural network models with image-based convolutional neural network models to generate comprehensive representations of cross-entity relationship graph data objects, augmenting predictive inferences and overcoming limitations of existing models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If graph convolutional neural network models are used to process complex graph data objects, then the model can handle graph-structured data, but it fails to accurately capture feature data related to node attributes, edge attributes, and edge weights
Solution Approach 1:
The patent combines graph convolutional neural networks with image-based convolutional neural networks into a hybrid framework. The graph CNN processes graph-structured data while the image CNN processes the generated graph image data object, allowing the system to capture both structural relationships and visual patterns in the data, thereby achieving comprehensive feature processing while maintaining high prediction accuracy
2Adaptability or versatility
If existing graph convolutional neural network models are used, then the processing approach is simple, but the model cannot holistically process interrelations of graph-based feature sources
Solution Approach 1:
The hybrid framework is segmented into distinct functional modules: a graph CNN module for processing graph data, an image generation module for creating graph image data objects, and an image CNN module for processing the visual representation. This segmentation allows each module to specialize in specific tasks while working together to achieve holistic processing of graph-based feature sources
3Measurement precision
If comprehensive representations are generated using hybrid models, then predictive accuracy improves, but computational operations and training data requirements increase
Solution Approach 1:
The framework performs preliminary action by generating a comprehensive representation of the graph data object before the main prediction task. The graph CNN and image CNN both process the input data to create intermediate representations that are then combined, preparing the data in a way that enhances subsequent prediction accuracy while distributing the computational workload across multiple specialized processors
Data Source
AI summary
Various embodiments of the present invention disclose techniques for determining a graph-based prediction based at least in part on a cross-entity relationship graph data object and using a hybrid graph-based processing machine learning framework. In some embodiments, the hybrid graph-based prediction machine learning framework is configured to generate the graph-based prediction based at least in part on a comprehensive representation of the cross-entity relationship graph data object that is generated based at least in part on output data of a graph convolutional neural machine learning model and an image-based graph convolutional neural network machine learning model.