connect

connect(pre, post, weights, bias=None, sparseness=1.0)

Connects two layers with a (sparse) weight matrix and optionally a bias vector.

Parameters

Name	Type	Description	Default
pre	Layer	input layer.	required
post	Layer	output layer.	required
weights	float \| RandomDistribution	float or `RandomDistribution` to create the weight matrix.	required
bias	float \| RandomDistribution	bias per post neuron. If `None` or `False`, no bias is used. Otherwise, can be a float or `RandomDistribution`.	`None`
sparseness	float	density of the weight matrix.	`1.0`

Name	Type	Description
	Projection	a `DenseProjection` or `SparseProjection` instance.