Struct FeatureHasher 

pub struct FeatureHasher { /* private fields */ }

Feature hashing (hashing trick) preprocessor.

Maps an arbitrary-length input feature vector to a fixed-size output of n_buckets dimensions. For each input feature at index i:

1. A bucket is chosen: bucket = hash(i, BUCKET_SEED) % n_buckets.
2. A sign is chosen: sign = if hash(i, SIGN_SEED) & 1 == 0 { +1 } else { -1 }.
3. The output accumulates: output[bucket] += sign * features[i].

The sign-flipping ensures that the inner product of two hashed vectors is an unbiased estimator of the original inner product (Weinberger et al., 2009).

Because the transform is purely deterministic and stateless, both update_and_transform and transform behave identically, and reset is a no-op.

Implementations

impl FeatureHasher

pub fn new(n_buckets: usize) -> Self

Create a new feature hasher with the given number of output buckets.

Panics

Panics if n_buckets is zero.

use irithyll::preprocessing::FeatureHasher;
let hasher = FeatureHasher::new(16);
assert_eq!(hasher.n_buckets(), 16);

pub fn n_buckets(&self) -> usize

Number of output buckets (output dimensionality).

Trait Implementations

impl Clone for FeatureHasher

fn clone(&self) -> FeatureHasher

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for FeatureHasher

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

Formats the value using the given formatter.

impl StreamingPreprocessor for FeatureHasher

fn update_and_transform(&mut self, features: &[f64]) -> Vec<f64>

Update internal statistics from this sample and return transformed features.

fn transform(&self, features: &[f64]) -> Vec<f64>

Transform features using current statistics without updating them.

fn output_dim(&self) -> Option<usize>

Number of output features, or None if unknown until the first sample.

fn reset(&mut self)

Reset to initial (untrained) state.