Package jline.api.mam

Class Mamap2m_fit_fb_multiclass

java.lang.Object

jline.api.mam.Mamap2m_fit_fb_multiclass

public final class Mamap2m_fit_fb_multiclass
extends Object

Port of m3a mamap2m_fit_fb_multiclass.m: marks a second-order acyclic MAP in canonical form with m classes so that the class probabilities are matched exactly and the per-class forward and backward moments as closely as the form allows.

The marking probabilities q(j,c) of the three D1 entries are affine in the fitted moments, q(j,c) = fF(c) q_f(j,c) + fB(c) q_b(j,c) + q_0(j,c), so the fit is a strictly convex quadratic program, solved exactly by M3a_qp as MATLAB solves it with its Goldfarb-Idnani QP. Degenerate forms, where only F, only B or nothing can be fitted, take the reference's dedicated branches.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static final class	Mamap2m_fit_fb_multiclass.FitResult	Fitted MMAP with the forward and backward moments it realizes.

Method Summary

Modifier and Type	Method	Description
static Mamap2m_fit_fb_multiclass.FitResult	mamap2m_fit_fb_multiclass(MatrixCell map, double[] p, double[] F, double[] B)
static Mamap2m_fit_fb_multiclass.FitResult	mamap2m_fit_fb_multiclass(MatrixCell map, double[] p, double[] F, double[] B, double[] classWeights, double[] fbWeights)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

mamap2m_fit_fb_multiclass

public static Mamap2m_fit_fb_multiclass.FitResult mamap2m_fit_fb_multiclass(MatrixCell map,
 double[] p,
 double[] F,
 double[] B)

mamap2m_fit_fb_multiclass

public static Mamap2m_fit_fb_multiclass.FitResult mamap2m_fit_fb_multiclass(MatrixCell map,
 double[] p,
 double[] F,
 double[] B,
 double[] classWeights,
 double[] fbWeights)

Parameters:

map - second-order acyclic MAP in canonical form 1 (gamma > 0) or 2 (gamma < 0)

p - class probabilities

F - target first-order forward moments

B - target first-order backward moments

classWeights - per-class weights (null for ones)

fbWeights - weights of the forward and backward moments (null for ones)