Parallel Graph Matching Using Data-Parallel Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current graph matching techniques are inefficient, especially as data sets increase in size, and are limited by serial methodologies, leading to performance bottlenecks.
Innovation Solution
The implementation of a system and method for optimized graph matching using data-parallel processing, including one-phase and two-phase handshaking methods, and algebraic multigrid techniques on a parallel architecture such as a GPU, to enhance efficiency and scalability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If serial methodologies are used for graph matching, then implementation simplicity is maintained, but processing efficiency and scalability deteriorate as data sets increase in size
Solution Approach 1:
The patent segments the graph matching process into independent vertex operations that can be executed in parallel. Each vertex independently selects neighbors based on predefined criteria, and these selections are processed simultaneously across multiple vertices, transforming a serial algorithm into a parallel one that scales with data size
Solution Approach 2:
The patent transitions from sequential time-based processing to spatial parallelism by mapping vertex operations onto a parallel architecture. This dimensional shift from one-dimensional serial execution to multi-dimensional parallel processing enables simultaneous computation across multiple vertices, dramatically improving throughput for large graphs
2Loss of time
If current graph matching implementations are used, then algorithm simplicity is preserved, but processing time increases significantly with larger data sets
Solution Approach 1:
The patent performs preliminary actions by pre-establishing selection criteria and neighbor relationships before the main matching process. Vertices pre-identify potential neighbors based on graph properties, allowing the actual matching decision to be made rapidly in parallel without time-consuming computations during the matching phase
Solution Approach 2:
The patent ensures continuous useful action by implementing iterative handshaking protocols where vertices continuously select and exchange neighbor information until convergence. This continuous parallel operation eliminates idle time and ensures maximum throughput, with each iteration contributing to the final matching result without interruption
Data Source
AI summary
A system, method, and computer program product are provided for performing graph aggregation. In use, a graph with a plurality of vertices and a plurality of edges is identified. Additionally, graph matching is performed on the vertices and edges of the graph by computing a graph matching, wherein the performance of the graph matching is optimized.