pyhgf.utils.weight_initialisation.he_init#

pyhgf.utils.weight_initialisation.he_init(n_parents, n_children, seed=None)[source]#

He / Kaiming normal initialisation.

Draws weights from \(\\mathcal{N}(0, \\sigma^2)\) where \(\\sigma = \\sqrt{2 / n_{\\text{parents}}}\). Designed for layers followed by ReLU activations.

Parameters:

  • n_parents (int) – Number of parent (input) nodes — fan-in.

  • n_children (int) – Number of child (output) nodes — fan-out.

  • seed (int | None) – Optional random seed for reproducibility.

Returns:

Weight vector of length n_parents * n_children.

Return type:

numpy.ndarray