ipfrs-tensorlogic 0.2.0

Zero-copy tensor operations and logic programming for content-addressed storage
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

//! EnsembleLearner — production-quality ensemble learning system.
//!
//! Implements Bagging, AdaBoost, Gradient Boosting, Random Forest, and Stacking
//! strategies over decision stumps and perceptron base models.
//!
//! # Design
//!
//! - Bagging: bootstrap samples, parallel independent base models, majority/average vote.
//! - AdaBoost: sequential stump fitting, exponential weight update (`alpha`), sample
//!   re-weighting after each round.
//! - Gradient Boosting: pseudo-residual fitting, shrinkage via `learning_rate`.
//! - Random Forest: bagging + feature sub-sampling per stump (sqrt(n_features)).
//! - Stacking: train diverse base models and a linear meta-learner on their outputs.
//! - All operations use `xorshift64` for bootstrap/sub-sampling — no `rand` crate.
//! - Training history is kept in a `VecDeque<ElTrainingRecord>` capped at 100 entries.
//! - No `unwrap()` anywhere — all fallible operations use `?` / `ok_or`.

pub mod ellearnerconfig_traits;
pub mod elmethod_traits;
pub mod functions;
pub mod type_aliases;
pub mod types;

// Re-export all types
pub use type_aliases::*;
pub use types::*;

#[cfg(test)]
mod tests;