Artifacts for "Common Evaluation Pitfalls in Touch-Based Authentication Systems".
The data used in this paper is available on ORA.
The features.zip
file contains the features.csv
which is the only file needed to recreate the experiments presented in the paper.
If you would like to perform the feature extraction yourself, the data_files.zip
folder contains all the necessary files with the following structure:
├── data_files
│ ├── UUID_X1 # User ID
│ │ │── SESSION_Y1
│ │ │ │── scroll # Social media task
│ │ │ │ │── REPETITON_W1
│ │ │ │ │ │── accelerometer_data
│ │ │ │ │ │ └── ###-###-###.csv
│ │ │ │ │ │── gyroscope_data
│ │ │ │ │ │ └── ###-###-###.csv
│ │ │ │ │ └── touch_data
│ │ │ │ │ └── ###-###-###.csv
| | | | └── REPETITON_W2
| | | | └── ...
│ │ │ └── swipe # Image gallery task
│ │ │ │── REPETITON_Z1
│ │ │ │ │── accelerometer_data
│ │ │ │ │ └── ###-###-###.csv
│ │ │ │ │── gyroscope_data
│ │ │ │ │ └── ###-###-###.csv
│ │ │ │ └── touch_data
│ │ │ │ └── ###-###-###.csv
│ │ | └── REPETITON_Z2
│ │ | └── ...
│ │ |
│ │ └── SESSION_Y2
│ │ └── ...
│ └── UUID_X2
│ │ └── ..
│ └── ...
To use this repository, you will also need docker
. Then complete the following steps:
git clone https://github.com/ssloxford/evaluation-pitfalls-touch.git
cd evaluation-pitfalls-touch/data
wget 'https://ora.ox.ac.uk/objects/uuid:5f1abaa7-52a4-430b-9208-128d9f1832fd/download_file?safe_filename=features.zip&type_of_work=Dataset' -O 'features.zip'
- (Optional) If you would like to generate the feature.csv yourself run:
wget 'https://ora.ox.ac.uk/objects/uuid:5f1abaa7-52a4-430b-9208-128d9f1832fd/download_file?safe_filename=data_files.zip&type_of_work=Dataset' -O 'data_files.zip'
instead.
- (Optional) If you would like to generate the feature.csv yourself run:
unzip features.zip
- (Optional)
unzip data_files.zip
- (Optional)
cd .. && docker build -t evaluation-pitfalls-touch .
- To start and connect to the docker container use:
docker run -it -v $(pwd)/data:/touch/data --name evaluation-pitfalls-touch evaluation-pitfalls-touch
- (Optional)
cd data && python3 feature_extraction.py && cd ..
- (Optional)
- You can now run the experiments - for instance
cd experiments && python3 p2_single_phone_model.py
We also share the source code of the iOS application in the that was developed for data collection purposes. This version can be found in the ios-app
folder but is not immediately functional as a back-end server is needed to make it work. However, it can be used as a reference or made usable with minor modifications.
This work was generously supported by a grant from Mastercard and by the Engineering and Physical Sciences Research Council [grant numbers EP/N509711/1, EP/P00881X/1].