Enabling hyperparameter optimization in sequential autoencoders for spiking neural data

We develop and test two potential solutions: an alternate validation method ("sample validation") and a novel regularization method ("coordinated dropout").