The code to accompany our technical report on aircraft noise monitoring.
The raw data can be found in two csv files the data subdir, namely
oml_final.csv and 400_community_events.csv. I read these into event2d
objects where the 1/3 octave data is available as event.rawmat. See
preprocess.py for an example of this.
TensorFlow > 1.0.0, Python 2, numpy, matplotlib
If you would like to run training and see results on a hold-out set,
navigate to the classifier subdir and run
make cvconvnet
This will train on 9/10 folds about 50 times and log the results.
Please contact Nicholas Heller at helle246@umn.edu if you have any questions about this work.