lqr_evp: EM Baggenstoss: lqr_evp

View source: R/lqr_evp.R

lqr_evpR Documentation

EM Baggenstoss: lqr_evp

Description

Helper function for creating classed based on Bayes decision with Expectation Maximization. Returns either total log-likelihood of Gaussian mixtures or separate log likelihoods for all modes.

Usage

lqr_evp(Parm, Data, Flag)

Arguments

Parm

Nested list with parameters for GMM. Features carrying permanent values. Features$name [1:d] String vector with feature names. Features$min_std [1:NMODE] Vector of covariance constraints. Modes carrying modifyable values. Modes$cholesky_covar [d*NMODE, d] Numerical matrix with NMODE many square matrices stacked vertically with the covariance matrix. Modes$mean [1:NMODE, d] Numerical matrix with nmode different means and d feature dimensions. Modes$weight [1, 1:NMODE] Numerical matrix with weights for each mean.

Data

[1:n,1:d] Numerical matrix with normalized data. N samples with DIM feature dimensions.

Flag

Boolean flag. flag==1, returns log mode PDF in columns, that is the output of lqr_eval only, not including mixing weights for each column. Flag=0: Computes the total log-PDF for normalized data in a single output column (i.e. the log of the weighted sum of the rows of the flag=1 output) plus the Jacobian (it outputs the PDF value with respect to the raw unnormalized data by taking into account the jacobian of the normalization operation). In this case outputs ModePDFs and PDF are derived.

Value

List with

lg

[1:n] Numerical vector with LogPDF

ModePDFs

[1:n,1:L] Numerical matrix with PDFs for all single modes.

PDF

[1:n] Numerical vector with combined Gaussian as weighthed sum of PDF.

Author(s)

Quirin Stier

References

Baggenstoss, Paul M., and T. E. Luginbuhl.: An EM algorithm for joint model estimation. IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No.99CH36258), Phoenix, AZ, USA, 1999, pp. 1825-1828 vol.4, IEEE, doi:10.1109/ICASSP.1999.758276, 1999.


Mthrun/AdaptGauss2D documentation built on July 19, 2022, 3:11 a.m.