Module gnn 

Graph Neural Network layers for learning on graph-structured data.

This module provides GNN layers commonly used for:

• AST/CFG analysis in transpilers (depyler, ruchy, bashrs)
• Code structure understanding
• Error pattern detection in CITL

Implemented Layers

• GCNConv - Graph Convolutional Network (Kipf & Welling, 2017)
• GATConv - Graph Attention Network (Veličković et al., 2018)
• SAGEConv - GraphSAGE (Hamilton et al., 2017)

Example

use aprender::nn::gnn::{GCNConv, GATConv, SAGEConv, AdjacencyMatrix};

// Create adjacency matrix for a simple graph
let adj = AdjacencyMatrix::from_edge_index(&[[0, 1], [1, 2], [2, 0]], 3);

// GCN layer
let gcn = GCNConv::new(64, 32);
let x = Tensor::new(&vec![0.0; 3 * 64], &[3, 64]);  // 3 nodes, 64 features
let out = gcn.forward(&x, &adj);  // [3, 32]

References

• Kipf, T. N., & Welling, M. (2017). Semi-Supervised Classification with Graph Convolutional Networks. ICLR.
• Veličković, P., et al. (2018). Graph Attention Networks. ICLR.
• Hamilton, W. L., et al. (2017). Inductive Representation Learning on Large Graphs. NeurIPS.

Structs

AdjacencyMatrix

Adjacency matrix representation for GNN operations.

GATConv

Graph Attention Network layer (Veličković et al., 2018).

GCNConv

Graph Convolutional Network layer (Kipf & Welling, 2017).

SAGEConv

GraphSAGE convolutional layer (Hamilton et al., 2017).

Enums

SAGEAggregation

Aggregation method for GraphSAGE.

Traits

MessagePassing

Message Passing Neural Network base trait.