Struct sparrow::scanner::booster::learner::Learner

pub struct Learner {
    pub rho_gamma: f32,
    pub expand_node: usize,
    pub total_count: usize,
    // some fields omitted
}

Statisitics of all weak rules that are being evaluated. The objective of Learner is to find a weak rule that satisfies the condition of the stopping rule.

Fields

rho_gamma: f32

expand_node: usize

total_count: usize

Methods

impl Learner

pub fn new(
    min_gamma: f32,
    default_gamma: f32,
    num_features: usize,
    bins: Vec<Bins>,
    num_splits: usize
) -> Learner

Create a Learner that search for valid weak rules. default_gamma is the initial value of the edge gamma. bins is vectors of the all thresholds on all candidate features for generating weak rules.

pub fn reset(&mut self)

Reset the statistics of all candidate weak rules (except gamma, because the advantage of the root node is not likely to increase) Trigger when the model or the gamma is changed

pub fn get_max_empirical_ratio_tree_node(&self) -> TreeNode

pub fn is_gamma_significant(&self) -> bool

pub fn update(
    &mut self,
    tree: &Tree,
    data: &[(LabeledData<u8, i8>, (f32, f32, usize, usize))]
) -> Option<TreeNode>

Update the statistics of all candidate weak rules using current batch of training examples.

pub fn set_gamma(&mut self, gamma: f32) -> bool

pub fn set_expand_node(&mut self, expand_node: usize) -> bool

Trait Implementations

impl<'de> Deserialize<'de> for Learner

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error> where
    __D: Deserializer<'de>, 

Deserialize this value from the given Serde deserializer.

impl Serialize for Learner

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error> where
    __S: Serializer, 

Serialize this value into the given Serde serializer.