Single-Solution Heuristic for Data Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing clustering algorithms face challenges in achieving superior solution quality and efficiency, particularly in terms of memory usage and the number of evaluations required to reach optimal values, especially when dealing with limited computational resources.
Innovation Solution
A single-solution-based heuristic method that combines features of Genetic Algorithms and Simulated Evolution, utilizing a greedy algorithm to select initial centroids and then iteratively updating solutions through mutation operations to maximize cluster separation and compactness, as measured by validity indices like Calinski Harabasz and Dunn indices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If population-based evolutionary algorithms are used to solve clustering, then solution quality can be improved, but memory consumption increases significantly
Solution Approach 1:
The patent extracts the essential evolutionary search mechanism from population-based algorithms and implements it in a single-solution framework. By taking out the core mutation and selection operations while removing the population structure, the algorithm achieves good solution quality with minimal memory usage, as it only needs to store one solution at a time rather than a population of solutions.
Solution Approach 2:
The single-solution heuristic serves itself by using its own current solution as the basis for generating new solutions through mutation. The algorithm improves its single solution iteratively without needing external population members, thereby reducing memory requirements while maintaining evolutionary search capabilities.
2Device complexity
If conventional heuristics are used for clustering, then computational simplicity is maintained, but the number of evaluations required to reach optimal values increases
Solution Approach 1:
The algorithm performs preliminary action by using a greedy initialization phase to quickly establish a good starting solution before entering the iterative improvement phase. This preliminary setup reduces the number of subsequent evaluations needed to reach optimal values, as the search starts from a more promising position rather than random initialization.
Solution Approach 2:
The patent implements continuous useful action through its iterative improvement mechanism that systematically explores the solution space. The algorithm continuously applies mutation operations and accepts improvements, maintaining a steady progression toward optimality without unnecessary computational interruptions or redundant evaluations.
3Reliability
If more evaluations are performed to reach optimal clustering results, then solution quality improves, but computational time increases
Solution Approach 1:
The algorithm applies dynamics by adjusting its search behavior based on the current state. The mutation rate and acceptance criteria dynamically adapt during the iterative process, allowing the algorithm to efficiently balance exploration and exploitation. This dynamic adjustment enables the algorithm to reach high solution quality with fewer evaluations, reducing computational time.
Data Source
AI summary
A heuristic method for the data clustering problem includes a greedy algorithm, which selects the centroids of well-separated clusters, and a single-solution-based heuristic, which performs clustering with the objective of optimizing a cluster validity index. The attributes of the centroids of clusters are genes and a fitness is determined for each gene. A mutation operation changes the values of the genes of low fitness values and creates a mutant solution. The values of genes of the mutant can replace the values of the genes of the parent if they are better than the genes of the parent. Diversification in the search process is maintained by allowing, with a small probability, the genes of the mutant to replace the genes of the parent even if they are not better than them.