amrElmTrain: Creates a model for AMR-ELM.

In rladeira/amrElm: Affinity Matrix Regularized Extreme Learning Machine

Description

Creates a model for AMR-ELM.

Usage

amrElmTrain(l, XTrain, YTrain, affinity = "cosine")

Arguments

l	the number of hidden neurons
XTrain	training data, numerical with zero mean and unit variance and patterns in the lines, attributes in the columns
YTrain	training data labels (binary, -1 and +1)
affinity	- only cosine implemented

Value

the amrElm model for supervised problems, with: Z: hidden layer weights H: hidden layer output W: output layer weights affinity: the affinity used to generate the model (e.g.: cosine affinity) dataTrain: training data for generating affinity matrix.

Examples

## Not run: 
library(amrElm)

data(heart)

data <- heart$data
labels <- heart$labels

l <- 500
N <- nrow(data)

randomPatterns <- seq(N)
data <- data[randomPatterns,]
labels <- labels[randomPatterns]

nTrain <- floor(2*N/3)
nTest <- N - nTrain

data <- data[randomPatterns,]
labels <- labels[randomPatterns]

XTrain <- data[1:nTrain,]
XTest <- data[(nTrain+1):N,]

YTrain <- labels[1:nTrain]

model <- amrElmTrain(l,XTrain,YTrain)

testOutput <- amrElmTest(XTest, model)

## End(Not run)

rladeira/amrElm documentation built on May 27, 2019, 9:17 a.m.