xPACS

Explaining Anomalies in Groups with Characterizing Subspace Rules. The work is accepted for publication and will appear in the proceedings of Data Mining and Knowledge Discovery Journal as part of the special issue: Journal Track of ECMLPKDD 2018.

Website

http://cmuxpacs.github.io/

Requirements

• Requires Matlab and R.
• Requires the following matlab packages:

cvx 

• Requires the following R packages:

data.table dplyr ggplot2 ROCR

File Structure

1. The data/ contains the public datasets used in the paper and src/ contains the code for xPACS as well as the other baselines in each folder.
2. The pre processed features and labels are stored in csv format in the data/ folder.
3. The full.sh script runs the end to end xPACS model for a dataset.

Contact

• Meghanath M Y (mmacha@cmu.edu)