PickGcnLayer<T>

Layers · AiDotNet.NeuralNetworks.Layers

One graph-convolution step of PICK’s GLCN (Yu et al., ICPR 2020; reference GCNLayer) over node-edge-node triplets: H_ij = ReLU(x_i W_vi + x_j W_vj + alpha_ij + b), x_i' = ReLU((sum_j A_ij H_ij) W_node), alpha_ij' = ReLU(H_ij W_alpha).

Inheritance: LayerBase<T> → PickGcnLayer<T>

Implements: IShapeContract

For Beginners

Each text box updates itself from every other box, weighted by how strongly the learned graph connects them and by where the boxes sit relative to each other.

How It Works

The single-input Forward takes nodes [N, D] and uses zero relation embeddings with a uniform adjacency. PICK drives Forward with its relation embedding and the learned soft adjacency.

Constructors

Constructor Summary
PickGcnLayer(int inDim, int outDim) Creates the layer.

Properties

Property Summary
bool SupportsTraining { get; }

Methods

Method Summary
Dictionary<string, string> GetParameterRoles() Returns parameter roles for per-role optimizer configuration (e.g., weight decay exemption for biases).
IReadOnlyList<OutputAxisContract> OutputAxesFor(int inputRank)
IReadOnlyList<Tensor<T>> GetTrainableParameters() Returns all trainable parameter tensors marked with [TrainableParameter].
void ResetState()
void SetTrainableParameters(IReadOnlyList<Tensor<T>> parameters) Replaces trainable parameter tensors (e.g., with ParameterBuffer views).
void ZeroGrad() Clears all gradient fields discovered by convention ({paramName}Gradient).