Node Centrality Calculation Using Matrix Exponential Approximation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calculating node centralities in large and complex networks are computationally expensive and inefficient, requiring significant computational resources and generating biases when sampling nodes to reduce complexity.
Innovation Solution
The method approximates the product of a matrix exponential and a random probe vector of an adjacency matrix using a stochastic estimator and mixed precision low complexity methods, allowing for the quick identification of the most central nodes and reducing computational cost and memory footprint.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing methods are used to calculate node centralities in large networks, then computational accuracy is maintained, but computational cost and time requirements become excessively high
Solution Approach 1:
The patent uses stochastic estimation to create a simplified computational model that copies only the essential characteristics of the full network analysis. Instead of computing exact node centralities for all nodes, it uses random probe vectors and matrix exponential approximations to estimate centralities for a subset of nodes, achieving speedup while maintaining sufficient accuracy for identifying important nodes.
Solution Approach 2:
The patent changes the computational parameters by using mixed precision arithmetic and approximate methods. It computes matrix exponential products using lower precision floating point operations and stochastic sampling, reducing the computational burden while maintaining the essential information needed for centrality analysis. This allows processing of large networks that would be intractable with exact methods.
2Loss of time
If sampling methods are used to reduce computational complexity, then processing time is reduced, but measurement precision and reliability deteriorate due to biases
Solution Approach 1:
The patent implements feedback mechanisms through iterative stochastic estimation. It uses random probe vectors to sample the network structure and adjusts the estimation process based on the results. The matrix exponential computation provides feedback about the network's global structure, allowing the method to refine its centrality estimates and reduce biases introduced by sampling.
Solution Approach 2:
The patent introduces random probe vectors as intermediaries between the network structure and the centrality measurements. These probe vectors serve as mediators that capture essential network characteristics without requiring direct analysis of all nodes and edges. The stochastic estimator uses these intermediaries to bridge the gap between computational efficiency and measurement accuracy.
3Measurement precision
If full network analysis is performed, then measurement precision is maintained, but memory requirements and device complexity increase significantly
Solution Approach 1:
The patent extracts only the essential information needed for centrality analysis from the full network. It uses random probe vectors to sample and extract key structural characteristics, then computes matrix exponential products only for these extracted samples rather than the entire network. This extraction approach maintains measurement precision for identifying central nodes while dramatically reducing memory and computational resource requirements.
Data Source
AI summary
Embodiments related to calculating node centralities in large and complex networks and graphs. An aspect includes approximating a product of a matrix exponential and a random probe vector of an adjacency matrix, wherein the adjacency matrix represents a graph. A diagonal of the adjacency matrix is computed based on the product of the matrix exponential and the random probe vector. The node centralities are then calculated based on the computed diagonal until a designated number of central nodes has been detected according to embodiments.