Segmentation based anomaly detection technique as described by Chan and Mahoney 2005.

Assign anomaly score to each test time series based on how well it fits the FSA learnt from the training data. The output is a single anomaly score for each test time series.
Usage:
./BOX -i -trainfilename -t testfilename -k numboxes -r numrandomrestarts -o outputfilename

-i      input training file name
-t      input testing file name
-o      output file
-k      number of boxes
-r      number of random restarts
