robust_weights: Compute weights

In JhZhang-1999/Rcvxclustr: Robust convex clustering: How does the fusion penalty enhance robustness?

Compute weights

Description

robust_weights computes the robust weights given a data matrix X, a scale parameter zeta and a parameter that controls the weights delta. Namely, the lth weight w[l] is given by

w[l] = exp(-ζ{∑_{j\in D_{1}}(X_{i'j}-X_{ij})^2+∑_{j\in D_{2}}{δ}^2}) , where the lth pair of nodes is (\code{i},\code{i'}) and \code{D1={j:|X_{ij}-X_{i'j}|<delta}}, \code{D2={j:|X_{ij}-X_{i'j}|>delta}}.

Usage

robust_weights(X, delta, zeta)

Arguments

X	The data matrix to be clustered. The rows are observations, and the columns are features.
delta	The nonnegative parameter that controls the scale of robust weights when there is outlier(s) in the data.
zeta	The nonnegative parameter that controls the scale of robust weights.

Value

A vector wt of weights for robust convex clustering.

Author(s)

Chenyu Liu, Qiang Sun, Kean Ming Tan

JhZhang-1999/Rcvxclustr documentation built on Aug. 18, 2022, 9:25 a.m.