Struct HypergraphAttentionNetwork 

pub struct HypergraphAttentionNetwork {
    pub layers: Vec<HypergraphAttentionLayer>,
    pub in_dim: usize,
    pub config: HypergraphAttentionConfig,
    /* private fields */
}

Multi-layer Hypergraph Attention Network.

Fields

layers: Vec<HypergraphAttentionLayer>

Stacked hypergraph attention layers.

in_dim: usize

Input dimension.

config: HypergraphAttentionConfig

Configuration (of the first layer, representative).

Implementations

impl HypergraphAttentionNetwork

pub fn new( in_dim: usize, n_layers: usize, config: HypergraphAttentionConfig, ) -> Self

Create a multi-layer Hypergraph Attention Network.

Arguments

• in_dim: input node feature dimension
• n_layers: number of stacked attention layers
• config: configuration (shared across layers)

pub fn forward( &self, node_feats: &Array2<f64>, incidence_matrix: &Array2<f64>, ) -> Result<Array2<f64>>

Forward pass through all layers.

Arguments

• node_feats: initial node features [N × in_dim]
• incidence_matrix: H ∈ {0,1}^{N×M}

Returns

Final node features [N × in_dim].

Trait Implementations

impl Clone for HypergraphAttentionNetwork

fn clone(&self) -> HypergraphAttentionNetwork

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for HypergraphAttentionNetwork

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

Formats the value using the given formatter.