networks – Structural Components: Networks

class pybrain.structure.networks.Network(name=None, **args)

Bases: pybrain.structure.modules.module.Module, pybrain.structure.parametercontainer.ParameterContainer

Abstract class for linking different modules with connections.

activate(inpt)

Do one transformation of an input and return the result.

activateOnDataset(dataset)

Run the module’s forward pass on the given dataset unconditionally and return the output.

addConnection(c)

Add the given connection to the network.

addInputModule(m)

Add the given module to the network and mark it as an input module.

addModule(m)

Add the given module to the network.

addOutputModule(m)

Add the given module to the network and mark it as an output module.

reset()

Reset all component modules and the network.

sortModules()

Prepare the network for activation by sorting the internal datastructure.

Needs to be called before activation.

class pybrain.structure.networks.FeedForwardNetwork(*args, **kwargs)

Bases: pybrain.structure.networks.feedforward.FeedForwardNetworkComponent, pybrain.structure.networks.network.Network

FeedForwardNetworks are networks that do not work for sequential data. Every input is treated as independent of any previous or following inputs.

class pybrain.structure.networks.RecurrentNetwork(*args, **kwargs)

Bases: pybrain.structure.networks.recurrent.RecurrentNetworkComponent, pybrain.structure.networks.network.Network

Class that implements networks which can work with sequential data.

Until .reset() is called, the network keeps track of all previous inputs and thus allows the use of recurrent connections and layers that look back in time.

addRecurrentConnection(c)

Add a connection to the network and mark it as a recurrent one.