Simulator for online credit card transactions that supports multi-modal authentication.
Please see the corresponding paper (MultiMAuS.pdf) for a detailed description of the simulator. Citation:
Zintgraf, L. M.; Lopez-Rojas, E. A.; Roijers, D. M.; and Nowé, A. 2017. MultiMAuS: A Multi-Modal Authentication Simulator for Fraud Detection Research. In Affenzeller, M.; Bruzzone, A. G.; Jimenez, E.; Longo, F.; and Piera, M. A., eds., 29th European Modeling and Simulation Symp. (EMSS 2017), 360–370. Curran Associates, Inc.
The main script that should be executed for a simulation is
simulator/transaction_model.py
. It needs a dictionary of
parameters as input. The default parameters that
are used as input for the simulator can be found in
simulator.parameters.py
.
The scripts we used for the experiments in the paper are:
experiments/run_unimaus.py
and
experiments/run_multimaus.py
.
The simulator takes aggregated data as input which is obtained from a real (private) dataset of credit card transactions. This aggregated data is taken from ./data/simulator_input.
See readme file in data folder for more information.
We also provide features that can be used for learning (e.g., classification)
from the transaction logs. For more information, see data/features
.