Have a look at Article A Comprehensive Survey of Clustering Algorithms

Article Survey of Clustering Algorithms

Daer Faisal Bin Ashraf,

Indeed the use of heuristic methods can increase the performance of clustering, please take a look at links and attached files on subject.

Article Heuristics for minimizing the maximum within-clusters distance

Conference Paper k-Means Performance Improvements with Centroid Calculation H...

Article How much k-means can be improved by using better initializat...

Article Improving performance of node clustering in wireless sensor ...

Article Improving performance of node clustering in wireless sensor ...

Best regards

Faisal Bin Ashraf have a look here

Chapter Heuristic Clustering Algorithms

I think there is no metaheurstic that can fits all types of problems. each algorithm has its advantages and its limitations (drop into local optima or not, number of parameters to set, the rapidity, the search space representation...etc)

Dear Faisal,

This definitely depends on the clustering model that you are willing to solve.

If you mean the minimum-sum-of-squares clustering model (the one for which K-means is a natural local minimizer), you can check the following article for an exact method based on column generation (quite efficient for small and medium problems):

Article An improved column generation algorithm for minimum sum-of-s...

and the following article plus related references regarding modern metaheuristics for larger problems:

Article HG-means: A scalable hybrid genetic algorithm for minimum su...

Good luck with your research!

--Thibaut

Have a look at

Conference Paper Data Clustering Using Hybrid Genetic Algorithm with k-Means ...