This is the repo for the EWSN'22 paper "Sardino: Ultra-Fast Dynamic Ensemble for Secure Visual Sensing at Mobile Edge"

pip3 install -r requirements.txt

Download MNIST dataset and store it under folder data_m/. Organize the data by following hierachy.

/data_m
  /MNIST
    /processed
      test.pt
      training.pt
    /raw
      ...

Download notMNIST dataset and store it under data_nm/. Organize the data by following hierachy.

/data_nm
  /Test
    /A
    /B
    ...
    /J
  /Train
    /A
    /B
    ...
    /J

Download GTSRB dataset in which the images are resized to 32x32. Organize the data by following hierachy.

/traffic-signs-data
  test.p
  train.p
  valid.p

Download KUL BelgiumTS dataset. Preprocess the downloaded data with KUL_preprocess.py. Orgainize the data by following hierachy.

/KUL
  test_data.npy
  test_labels.npy
  train_data.npy
  train_labels.npy

python3 train_hypernet.py --cuda --dataset mnist

python3 train_hypernet.py --cuda --dataset gtsrb

python3 train_hypernet.py --cuda --dataset kul

python3 experiments.py --cuda