Prerequiste: make sure that Keras is installed and spearmint framwork works propoerly through running braninpy example as described in https://github.com/JasperSnoek/spearmint

To run Similarity Analysis example:

1. build the workload representation model:

cd source-code/similarity-analysis
python exec_features_representation_learner.py 

2. run example that analyzes similarity across the dataset workloads (the dataset is based on HiBench workloads https://github.com/Intel-bigdata/HiBench):

python analyze_similarity.py

To run SimTune example:

1. cd source-code
2. edit simtune/config.pb to point to the source workload (i.e. set the 'name' parameter to the file that contains the workload code) and the configuration paramters then start tuning using the following command:

python spearmint_sync.py simtune --method=MultiTaskEIOptChooser --method-args=\"task_num=0\"  --max-finished-jobs=15

3. To transfer this tuning knowledge over to a target workload: edit simtune/config.pb to point to the target workload then start tuning using the following command:

python spearmint_sync.py simtune --method=MultiTaskEIOptChooser --method-args=\"task_num=1\"  --max-finished-jobs=30