Skip to main content


Fig. 14 | Journal of Big Data

Fig. 14

From: Efficiency of random swap clustering

Fig. 14

Effect of the number of k-means iterations applied within RS. The value 2 is slightly better choice but the exact value is not critical. Only when iterated very long, high number of k-means iterations can become more beneficial with the image datasets. When running 45 min with Bridge, iterating k-means until convergence improved the SSE-value from 161.38 to 160.89 (improvement of 0.3%)

Back to article page