This contains code to apply deep CNN networks for LTE signal classification
Basic Info:
train_model.py is the main file for training the model.
core/ contains the code to introduce CNN layers
All data needs to be stored in folder data/
Before running create a folder to save checkpoints in and provide it as input
Sample Command:
python train_model.py -d lte_tdwaveform.csv -dL train_lbl.csv -v lte_tdwaveform_val.csv -t lte_tdwaveform_test.csv -tL test_lbl.csv -vL val_lbl.csv
Dependencies: 1. Python 2.7 distribution with numpy - Ideally use anaconda2
2. Theano
- gcc
- blas
- CUDA (optional)
- Follow instruction here for simple cpu install on windows
http://stackoverflow.com/questions/33687103/how-to-install-theano-on-anaconda-python-2-7-x64-on-windows
- Instructions here for adding blas and gpu support
https://github.com/Lasagne/Lasagne/wiki/From-Zero-to-Lasagne
http://computerscienceunveiled.blogspot.in/2015/08/installing-openblas-for-theano-on.html
3. Keras
- simple 'pip install keras'