Vector Clustering by Cosine Sorting for Fast Convergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing clustering algorithms are sensitive to initial placement of cluster centers and suffer from the curse of dimensionality, making them unsuitable for time-constrained scenarios such as cybersecurity threat analysis and telecommunications network reconfiguration.
Innovation Solution
A computer-implemented method that sorts vectors by cosine distance from an origin vector, divides the list into clusters based on directional differences, and iteratively refines the clustering solution by reassigned vectors to improve fitness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If gradient descent is used to find maxima or minima in clustering algorithms, then the algorithms can optimize cluster centers, but they become highly sensitive to initial placement of cluster centers and slow to converge
Solution Approach 1:
The patent applies preliminary action by sorting vectors by cosine distance from an origin vector before clustering. This pre-processing step organizes the data in a meaningful order that guides subsequent clustering operations, eliminating the need for random initial placement of cluster centers and avoiding the sensitivity issues of gradient descent methods.
Solution Approach 2:
The patent segments the sorted vector list into K ordered list fragments by positioning fragment boundaries at the top K distances. This segmentation approach divides the clustering problem into manageable segments that can be processed independently, achieving both accuracy and computational efficiency without requiring iterative optimization.
2Adaptability or versatility
If traditional clustering algorithms are used, then they can group data into categories, but they suffer from the curse of dimensionality making them slow to converge
Solution Approach 1:
The patent replaces the mechanical iterative optimization process of traditional clustering algorithms with a direct computational approach. By using cosine distance sorting and fragment boundary positioning, the method substitutes iterative mechanical adjustments with a single-pass computational procedure that is not affected by the curse of dimensionality.
Solution Approach 2:
The patent changes the fundamental parameters of the clustering approach by using cosine distance as the sorting criterion and defining clusters based on distance thresholds rather than iterative centroid updates. This parameter change transforms the algorithm from one that suffers from dimensional curse to one that scales efficiently with dimensionality.
3Reliability
If manual labeling of training data is performed, then machine learning models can be trained, but it is time-consuming
Solution Approach 1:
The patent applies self-service by enabling the clustering algorithm to automatically group vectors into meaningful categories without requiring manual labeling. The algorithm uses intrinsic properties of the data (cosine distances) to perform segmentation, making the system self-sufficient and eliminating the time-consuming manual labeling process while maintaining reliable clustering results.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
A computer-implemented method of clustering a plurality of vectors, the computer-implemented method comprising: sorting the plurality of vectors by cosine distance from an origin vector to generate an ordered circular list; and dividing the ordered circular list into a plurality of ordered list fragments, each ordered list fragment consisting of one or more vectors, thereby generating a solution in which the one or more vectors constituting each of the plurality of ordered list fragments are assigned to a respective one of a corresponding plurality of vector clusters.