EM: Elastic Map

In ZahraSajjad/ElasticMap: Elastic Map Fitting

Description

EM is function which fit parameters of elastic map. The main idea of this approach is to search map as solution of optimization problem.

Usage

EM(Data_Map, data, varargin)

Arguments

Data_Map	is the list of preprocessed data and initialized map variable that are obtaind with init function
data	The data is n-by-m matrix with m data points and n coordinates for each point (each column is one data point)
varargin	is a list of fitting options that can be one of following strings: type: can be hard, medium or soft. "hard" is hard map with bend = 1 and stretch = 1. "medium" is flexible map with bend = 0.7 and stretch = 0.7 and "soft" is soft map with bend = 0.5 and stretch = 0.5 stretch: is a positive numeric value which is the value of stretching modulo, "bend" is a positive numeric value which is the value of the bending modulo. weights: is n-by-1 vector of weights for data points. Weights must be nonnegative. Number_of_intervals: specifies the number of intervals to automatic interval calculation. Default value is 5. Ignored if "potential" is not specified. intshrinkage: is fraction of maximal distance from data points to original map which is used for intervals shrinkage. Default value is 1 (no shrinkage). Ignored if "potential" is not specified. potential: is majorant function for PQSQ. L2 distance without shrinkage is used if 'potential' is not specified.

Value

it is a List same as Data_Map variable that map.mapped is changed because of Elastic Map training.

Author(s)

Zahra Reazaei

Examples

# Data Points
data <- BreastCancer_Data

# Create Map Geometry
map <- rect2Dmap(10,10)

# initialize Map
Data_Map <- init(map, data,"Pci")

# Elastic Map Training
varargin <- list("stretch", 0.01,"bend", 0.1)
Data_Map <- EM(Data_Map, data, varargin)

ZahraSajjad/ElasticMap documentation built on July 26, 2020, 12:38 a.m.