rdkit.ML.Neural.NetNode module

Contains the class _NetNode_ which is used to represent nodes in neural nets

Network Architecture:

A tacit assumption in all of this stuff is that we’re dealing with feedforward networks.

The network itself is stored as a list of _NetNode_ objects. The list is ordered in the sense that nodes in earlier/later layers than a given node are guaranteed to come before/after that node in the list. This way we can easily generate the values of each node by moving sequentially through the list, we’re guaranteed that every input for a node has already been filled in.

Each node stores a list (_inputNodes_) of indices of its inputs in the main node list.

class rdkit.ML.Neural.NetNode.NetNode(nodeIndex, nodeList, inputNodes=None, weights=None, actFunc=<class 'rdkit.ML.Neural.ActFuncs.Sigmoid'>, actFuncParms=())

Bases: object

a node in a neural network

Eval(valVect)

Given a set of inputs (valVect), returns the output of this node

Arguments

• valVect: a list of inputs

Returns

the result of running the values in valVect through this node

GetInputs()

returns the input list

GetWeights()

returns the weight list

SetInputs(inputNodes)

Sets the input list

Arguments

• inputNodes: a list of _NetNode_s which are to be used as inputs

Note

If this _NetNode_ already has weights set and _inputNodes_ is a different length, this will bomb out with an assertion.

SetWeights(weights)

Sets the weight list

Arguments

• weights: a list of values which are to be used as weights

Note

If this _NetNode_ already has _inputNodes_ and _weights_ is a different length, this will bomb out with an assertion.