Advances, Systems and Applications
Table 5 Random parameter range of workload, resource and environmental conditions in the experiment
Parameter settings | Random value range | |
|---|---|---|
Training image data set variation | Number of training images | [1 ~ 30]*1000 |
Training image pixels | horizontal[4 ~ 64], vertical [4 ~ 64] | |
Image recognition model variation | Number of dense layers | 1–20 |
Number of neurons for each dense layer | 1–50 | |
Number of convolutional layers | 1–20 | |
Number of neurons for each convolutional layer | 1–50 | |
Activation function for each layer | softmax, relu, sigmoid | |
Resource amount range | Number of vCPUs allocated to VM | 1 ~ 8 |
Size of vMemory allocated to VM | 16 ~ 64GB | |
Environmental conditions range | CPU usage of host | 0 ~ 100% |
Memory usage of host | 0 ~ 100% |