• numpy
• TensorFlow
• scikit-optimize
• fanova to install, please follow the steps here

If you are using pip package manager, you can simply install all requirements via the following command(s):

python -m virtualenv .env -p python3 [optional]
source .env/bin/activate [optional]
pip3 install -r requirements.txt

1. You can get the preview of the SHL dataset (tar.gz files) from here. Make sure to put the downloaded files into ./data/ folder.
2. Run extract_data.sh script which will extract the dataset into ./generated/tmp/ folder.

In order to run Bayesian optimization, you can issue the following command:

python3 shl-nas.py --run bayesopt

Additionally, you can specify a subset of the data generators you want to apply Bayesian optimization on as follows:

python3 shl-nas.py --run bayesopt --position {bag|torso|hand|hips}

You can find a complete notebook showing the functional analysis of variance inside notebooks/ folder.

Similarly, you can train a single model by issuing the following command:

python3 shl-nas.py --run trainSingleModel [--position {bag|torso|hand|hips}]

• Massinissa Hamidi ([email protected])
• Aomar Osmani