Backprop Network

A backprop network is a feed forward network that can be trained using the backpropagation learning algorithm, which is perhaps the best known and most popular means of training neural networks. The Simbrain implementation (currently fairly bare bones) is described here

Backprop networks are created using Insert network > backprop. They can then be trained using the training dialog, and then either used by themselves or linked to other components. Once a backprop network has been trained data can be validated in it using the validation tab of the training dialog. Backprop can also be used separately from a backprop network, in scripts to custom train a set of weights. An example which steps through the process of creating and training a backprop network is in the examples page.

Creation Dialog

The backprop creation dialog allows you to specify the network topology and neuron types. This dialog is documented in feed forward docs.

Training

Training a network involves specifying input data, target data, and then running the algorithm. This process is handled in the training dialog. To open this dialog double click on the interaction box that is by default labelled "Backprop"

Once the network is trained to perform a particular input-output mapping, it can be linked to other networks or neurons in Simbrain, or simply used on its own, primarily with the validation tab of the training dialog.

Right Click Menu

Generic right-click items are described on the neuron group page.

Edit/Train Backprop: Opens edit dialog to edit and train the Backprop network.