admm_nmr_clf_top:

In riverKangg/NuclearNormClassifier: Nuclear Norm Classifier

Usage

  admm_nmr_clf_top(A, B, label, top_num)

Arguments

A	"A" is a train data set in the form of a three-dimensional array. This data set contains several face images used for algorithm learning.
B	"B" is an image of a face with occlusion.
label	"label" is a vector representing a label for train dataset A.
top_num	"top_num" represents the number of classification candidates. If "top_num" is 1, this function is the same as admm_nmr_clf.

Value

val_min[0]	The
val_top

Author(s)

Jisun Kang

References

https://ieeexplore.ieee.org/document/7420697

Examples

function (A, B, label, top_num)
{
    x <- ADMM_NMR_fit(A, B)
    A_x <- coef_img(A, x)
    label_unique <- unique(label)
    error_lst <- c()
    for (i in target_unique) {
        label_tf <- (label == label_unique)
        x_i <- x[label_tf]
        A_i <- A[, , label_tf]
        A_x_i <- coef_img(A_i, x_i)
        diff <- A_x - A_x_i
        error <- nuclear(diff)
        error_lst <- append(error_lst, error)
    }
    error_top <- sort(error_lst)[1:top_num]
    error_top_tf <- is.element(error_lst, error_top)
    error_min_tf <- (error_lst == min(error))
    val_min <- label[error_top_tf]
    return(val_min[0], val_top)
  }

riverKangg/NuclearNormClassifier documentation built on Jan. 1, 2020, 10:08 p.m.