/sdae

Stacked denoising autoencoders

Primary LanguagePython

Setup

Install PyTorch, NumPy, and other common packages if you don't have them already.

pip install -r requirements.txt

There are two datasets you'll need to download manually (see below). I suggest you create a data folder and unpack the relevant files into it. Later, you will be able to specify the dataset paths as command line arguments.

Olshausen Link
CUB-200-2011 Link

You can run one of the provided experiment scripts using the command ./experiments/xx.sh, where xx is the experiment number (e.g. 01). I have included a brief experiment description at the top of each script. You will need to change arguments such as olshausen_path and cub_folder which refer to dataset paths.

Associated Visuals

ofmfts