spamosaic.architectures.hg_lgcn

class spamosaic.architectures.hg_lgcn.HG_LGCN(*args: Any, **kwargs: Any)[source]

Bases: Module

Full Heterogeneous Graph LightGCN network with encoder-decoder architecture.

This module uses HG_LGCN_vanilla for graph feature encoding and a customizable decoder for reconstruction.

Parameters:

input_size (int) – Dimensionality of input node features.
output_size (int) – Dimension of the latent embedding.
K (int, default=8) – Number of LightGCN layers.
dec_l (int, default=1) – Number of layers in the decoder (1 = linear decoder).
hidden_size (int, default=512) – Size of hidden layer in the MLP head.
dropout (float, default=0.2) – Dropout rate used in MLP head.

forward(feature, edge_index, edge_weight, edge_type)[source]: Returns: - Latent embedding (L2-normalized). - Reconstructed feature from decoder.

forward(feature, edge_index, edge_weight, edge_type)[source]

Forward pass for HG_LGCN model.

Parameters:

feature (torch.Tensor) – Input node features.
edge_index (torch.LongTensor) – COO edge list.
edge_weight (torch.Tensor) – Edge weights.
edge_type (torch.Tensor) – Edge type indicators.

Returns:

Normalized latent representation (embedding).
Reconstructed feature tensor.

Return type:

Tuple[torch.Tensor, torch.Tensor]

class spamosaic.architectures.hg_lgcn.HG_LGCN_Conv(*args: Any, **kwargs: Any)[source]

Bases: MessagePassing

Heterogeneous Graph LightGCN convolution layer.

This layer handles intra-group and inter-group message passing separately, applying degree-normalized edge weights and combining both feature streams.

Inherits from:: torch_geometric.nn.conv.MessagePassing

forward(x, edge_index, edge_weight, edge_type)[source]: Executes the forward pass using intra- and inter-group edge types.

message(x_j, norm)[source]: Computes messages for each edge using normalized weights.

forward(x, edge_index, edge_weight, edge_type)[source]

Forward pass for heterogeneous LightGCN convolution.

Parameters:

x (torch.Tensor) – Input node features, shape (N, F).
edge_index (torch.LongTensor) – Edge list in COO format, shape (2, E).
edge_weight (torch.Tensor) – Edge weights, shape (E,).
edge_type (torch.Tensor) – Edge type indicator: 1 for intra-group, 0 for inter-group.

Returns:

Concatenated features from intra-group and inter-group aggregations.

Return type:

torch.Tensor

message(x_j, norm)[source]

class spamosaic.architectures.hg_lgcn.HG_LGCN_vanilla(*args: Any, **kwargs: Any)[source]

Bases: Module

Stacked HG_LGCN_Conv layers with layer-wise feature concatenation.

Parameters:: num_layers (int) – Number of HG_LGCN_Conv layers to stack.

forward(x, edge_index, edge_weight, edge_type)[source]: Executes stacked HG_LGCN_Conv layers and concatenates intermediate features.

forward(x, edge_index, edge_weight, edge_type)[source]

Forward pass of stacked HG_LGCN layers.

Parameters:

x (torch.Tensor) – Input node features.
edge_index (torch.LongTensor) – Edge list in COO format.
edge_weight (torch.Tensor) – Weights for each edge.
edge_type (torch.Tensor) – Indicator for intra/inter group edges.

Returns:

Concatenated node features from all layers.

Return type:

torch.Tensor

Classes

`spamosaic.architectures.hg_lgcn.HG_LGCN`	Full Heterogeneous Graph LightGCN network with encoder-decoder architecture.
`spamosaic.architectures.hg_lgcn.HG_LGCN_Conv`	Heterogeneous Graph LightGCN convolution layer.
`spamosaic.architectures.hg_lgcn.HG_LGCN_vanilla`	Stacked HG_LGCN_Conv layers with layer-wise feature concatenation.