olfactory (multi-trial GPLVM)
The code is for paper Learning a latent manifold of odor representations from neural responses in piriform cortex.
Run gen_syn_2d.ipynb to generate 2d simulated data.
Run demo1.ipynb and demo2.ipynb for multi-trial GPLVM fit to the simulated data. You will recover the latent and reconstruct the firing rates.
demo2.ipynb consists of demo1.ipynb and a second stage. The first stage (same as demo1.ipynb) estimates latent and model parameters with the naive model assumption. The second stage is a fine-tune of latent and model parameters with the user-specified model assumption.
More details can be found in the notebooks.
All required packages are included in olfactory.yml. You can install a conda env via
conda env create -f olfactory.yml