This is the implementaion of the inference developed in "Improving the Gaussian Process Sparse Spectrum Approximation by Representing Uncertainty in Frequency Inputs". Please see the paper for further details.
This code assumes that the Theano repository was cloned in the parent folder. To clone the repo, run:
git clone https://github.com/Theano/Theano.git
To run the example file, execute:
python VSSGP_example.py
Note that Theano will compile the code the first time the example is run, which might take some time.