Struct SAGEConv 

pub struct SAGEConv { /* private fields */ }

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

Implements the aggregation rule: h_v^(l+1) = σ(W · CONCAT(h_v^(l), AGG({h_u^(l) : u ∈ N(v)})))

Where AGG can be mean, max, sum, or LSTM aggregation.

Example

use aprender::nn::gnn::{SAGEConv, AdjacencyMatrix, SAGEAggregation};
use aprender::autograd::Tensor;

let sage = SAGEConv::new(64, 32).with_aggregation(SAGEAggregation::Mean);
let x = Tensor::new(&vec![0.1; 5 * 64], &[5, 64]);
let adj = AdjacencyMatrix::from_edge_index(&[[0, 1], [1, 2], [2, 3], [3, 4]], 5);
let out = sage.forward(&x, &adj);
assert_eq!(out.shape(), &[5, 32]);

Implementations

impl SAGEConv

pub fn new(in_features: usize, out_features: usize) -> SAGEConv

Create a new GraphSAGE layer.

Arguments

• in_features - Input feature dimension per node
• out_features - Output feature dimension per node

pub fn with_aggregation(self, agg: SAGEAggregation) -> SAGEConv

Set aggregation method.

pub fn with_normalize(self) -> SAGEConv

Enable L2 normalization of output.

pub fn without_root(self) -> SAGEConv

Disable root weight (self features).

pub fn without_bias(self) -> SAGEConv

Disable bias.

pub fn in_features(&self) -> usize

Get input feature dimension.

pub fn out_features(&self) -> usize

Get output feature dimension.

pub fn aggregation(&self) -> SAGEAggregation

Get aggregation method.

pub fn forward(&self, x: &Tensor, adj: &AdjacencyMatrix) -> Tensor

Forward pass with neighbor aggregation.

Arguments

• x - Node features [num_nodes, in_features]
• adj - Adjacency matrix

Returns

Output features [num_nodes, out_features]

Trait Implementations

impl Clone for SAGEConv

fn clone(&self) -> SAGEConv

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for SAGEConv

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.