Struct GraphGradientAttribution 

pub struct GraphGradientAttribution {
    pub gradients: HashMap<String, Tensor>,
    pub smooth_steps: usize,
}

Gradient-based attribution methods for graphs

Fields

gradients: HashMap<String, Tensor>

Store gradients for analysis

smooth_steps: usize

Smoothing parameter for integrated gradients

Implementations

impl GraphGradientAttribution

pub fn new() -> Self

pub fn integrated_gradients( &self, graph: &GraphData, _baseline_graph: &GraphData, _target_class: usize, ) -> Result<GraphData, Box<dyn Error>>

Compute integrated gradients for graph inputs

pub fn gradient_saliency( &self, graph: &GraphData, _target_output: &Tensor, ) -> Result<Tensor, Box<dyn Error>>

Compute gradient-based saliency maps for nodes

Trait Implementations

impl Clone for GraphGradientAttribution

fn clone(&self) -> GraphGradientAttribution

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GraphGradientAttribution

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

Formats the value using the given formatter.

impl Default for GraphGradientAttribution

fn default() -> Self

Returns the “default value” for a type.