Struct MaxNormRegularization 

pub struct MaxNormRegularization {
    pub max_norm: f64,
    pub lambda: f64,
    pub axis: usize,
}

MaxNorm constraint regularizer.

Constrains the norm of weight vectors to a maximum value. Useful for preventing exploding gradients and improving stability.

References

• Hinton et al.: “Improving neural networks by preventing co-adaptation”

Fields

max_norm: f64

Maximum allowed norm

lambda: f64

Regularization strength

axis: usize

Axis along which to compute norms (0 for rows, 1 for columns)

Implementations

impl MaxNormRegularization

pub fn new(max_norm: f64, lambda: f64, axis: usize) -> Self

Create a new max norm regularizer.

Trait Implementations

impl Clone for MaxNormRegularization

fn clone(&self) -> MaxNormRegularization

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MaxNormRegularization

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

Formats the value using the given formatter.

impl Default for MaxNormRegularization

fn default() -> Self

Returns the “default value” for a type.

impl Regularizer for MaxNormRegularization

fn compute_penalty( &self, parameters: &HashMap<String, Array<f64, Ix2>>, ) -> TrainResult<f64>

Compute the regularization penalty for given parameters.

fn compute_gradient( &self, parameters: &HashMap<String, Array<f64, Ix2>>, ) -> TrainResult<HashMap<String, Array<f64, Ix2>>>

Compute the gradient of the regularization penalty.