Table 3 Makespan ratio and percentage makespan reduction measured by running K-means and Random Forest algorithms on the HEPMASS and MNIST datasets
From: Block size estimation for data partitioning in HPC applications using machine learning techniques
Algorithm | Dataset name | Dataset rows | Dataset columns | Metric | Best time | Average time | Worst time |
|---|---|---|---|---|---|---|---|
K-means | HEPMASS | \(7\cdot 10^6\) | 27 | Makespan ratio | \(0.96 \pm 0.03\) | \(1.48 \pm 0.04\) | \(2.53 \pm 0.07\) |
Makespan red. | \(-3.80\% \pm 0.09\) | \(32.6\% \pm 0.05\) | \(60.5\% \pm 0.06\) | ||||
Random Forest | MNIST | \(6\cdot 10^4\) | 784 | Makespan ratio | \(1.00 \pm 0.01\) | \(1.27 \pm 0.03\) | \(1.65 \pm 0.03\) |
Makespan red. | \(0\% \pm 0.01\) | \(21.32\% \pm 0.04\) | \(39.51\% \pm 0.06\) |