Implementation of the paper "A scalable end-to-end Gaussian process adapter for irregularly sampled time series classification".
- Python 2.7
- Numpy
- Scipy
- Theano
- Lasagne
- Cython:
run the command
python setup.py build_ext --inplace
to compile Cython modules.
Train GP adapter with a convolutional net for 100 epochs on the sparsified uWave data:
python gpnet_fast.py -f data/UWaveGestureLibraryAll-1.pkl -e 100 -n cnn
Run python gpnet_fast.py -h
to learn the command line options.