Design Choices in Deep Learning (Web-interface)
This HTML page contains the two Shiny based Web-interfaces for designing deep networks i.e., choosing the best set of network sizes and learning parameters for a given choice ( user defined ) of hyper-parameters.
The first inteface shows the design choices in the fully-supervised case where unsupervised instances are unavailable.
The second one shows the design choices in the alternate case where unsupervised data is available.
The following Interfaces are currently being updated. Instead, please download the corresponding ShinyApps directly and run the apps in R/RStudio.
The Apps can be found here Full-supervised Design and
Pretraining + Fine-tuning Design respectively.