Adar : Adversarial Activity Recognition in Wearables

This repository has the code and data for the experiments we did in our paper Adar : Adversarial Activity Recognition in Wearables

Logistics

1. To run the provided code, you need the following tools:
   1. Python 3.6.8, the latest version of Python.
   2. TensorFlow version 1.12.0
   3. Numpy, Pandas and other tools.

The best way to get these tools is to use the Anaconda

Execution

• To run the experiments you can either use the provided pre-trained models and processed data. To do this just run the Python script file run_pretrained.py. This will run all the experiments using the models and data provided in the repository and store the results in the folder results/pretrained/.