water-vapor
commented
3 years ago Thank you for your interest in our package. Though you may see a Dataset class from some examples, it's just a simple utility class that (down)loads the dataset and converts to (batch_size, timestep, input_size) arrays. The learning algorithms will work with any numpy arrays of such shapes, including real-valued and spiking datasets (the actual effectiveness will depend on the algorithm). Since users may have custom preprocessing of these datasets, we may not prioritize adding them to the main package
Are DVS datasets going to be added in the future? Is it straightforward to train spiking datasets? any help would be appreciated.