LinkPredictionDecoder

Enums · AiDotNet.Enums

Defines how a link-prediction model scores a candidate edge from the two node embeddings at its ends.

For Beginners

The model turns every node into a list of numbers. To guess whether two nodes should be connected it has to turn that pair of lists into a single score. These are the ways it can do that — multiply them together, measure the angle between them, combine them element by element, or measure how far apart they are.

How It Works

Promoted out of LinkPredictionModel<T>, where it was nested. A type nested in a generic class is a distinct type per type argument, so the value could not be named on a non-generic options class.

Fields

Field Summary
LinkPredictionDecoder CosineSimilarity Cosine similarity — the dot product normalised by length, so only direction counts.
LinkPredictionDecoder Distance Distance between the embeddings, so nearby nodes score as likely neighbours.
LinkPredictionDecoder DotProduct Dot product of the two embeddings.
LinkPredictionDecoder Hadamard Element-wise (Hadamard) product, fed to a learned scoring layer.
int value__