anofox-ml-svm 0.1.0

//! # anofox-ml SVM
//!
//! Support Vector Machine classifiers for the anofox-ml machine learning library.
//!
//! This crate provides two SVM classifiers:
//!
//! - [`LinearSvc`] -- Linear Support Vector Classifier using hinge loss + L2
//!   regularization, solved via coordinate descent (similar to sklearn's
//!   `LinearSVC` with liblinear).
//! - [`Svc`] -- Support Vector Classifier with kernel support (linear, RBF,
//!   polynomial), solved via a simplified SMO algorithm.
//!
//! Both classifiers support binary and multi-class classification (via
//! one-vs-rest strategy).
//!
//! ## Example
//!
//! ```rust
//! use anofox_ml_core::{Fit, Predict};
//! use anofox_ml_svm::{LinearSvc, Svc, SvmKernel};
//! use ndarray::array;
//!
//! // Linear SVC
//! let x = array![[0.0, 0.0], [0.1, 0.1], [5.0, 5.0], [5.1, 5.1]];
//! let y = array![0.0, 0.0, 1.0, 1.0];
//!
//! let svc = LinearSvc::new().with_c(1.0);
//! let model = svc.fit(&x, &y).unwrap();
//! let preds = model.predict(&x).unwrap();
//!
//! // Kernel SVC with RBF
//! let svc = Svc::new()
//!     .with_kernel(SvmKernel::Rbf { gamma: 0.5 })
//!     .with_c(10.0);
//! let model = svc.fit(&x, &y).unwrap();
//! let preds = model.predict(&x).unwrap();
//! ```

mod kernel;
mod linear_svc;
mod linear_svr;
mod nu_svc;
mod nu_svr;
mod one_class_svm;
mod svc;
mod svr;

pub use kernel::SvmKernel;
pub use linear_svc::{FittedLinearSvc, LinearSvc};
pub use linear_svr::{FittedLinearSvr, LinearSvr};
pub use nu_svc::{FittedNuSvc, NuSvc};
pub use nu_svr::{FittedNuSvr, NuSvr};
pub use one_class_svm::{FittedOneClassSvm, OneClassSvm};
pub use svc::{FittedSvc, Svc};
pub use svr::{FittedSvr, Svr};