build_xnnet: Building xnnet

In bioinfo00/xnnet0: Explainable neural network classification

Description

This function builds xnnet using a labeled training set of transcriptional profiles, and a list of annotation libraries

Usage

build_xnnet(
  X_train,
  y_train,
  annotation_libraries,
  n_input_nodes = 3,
  n_hidden_nodes = 4,
  n_unassigned = 3,
  number = 10,
  min_decay = 0.1,
  max_decay = 1
)

Arguments

X_train	data matrix
y_train	binary labels
annotation_libraries	a list of annotation libraries. Each library is a list of gene sets gene sets
n_input_nodes	maximum number of input nodes per hidden nodes
n_hidden_nodes	number of hidden nodes in the network
n_unassigned	number of input genes not assigned to a specific gene set
number	number of bootstrap samples for network training
min_decay	minimum value of the decay parameter for regularization
max_decay	maximum value of the decay parameter for regularization

Value

A list containing an xnnet object for each annotation library

Examples

data("GSE37250") #load Tubercolosis dataset
data("annotation_libraries")
GSE37250_split = train_test_split(GSE37250$X, GSE37250$y)
xnnet = build_xnnet(X_train = GSE37250_split$X_train, y_train = GSE37250_split$y_train,
annotation_libraries = annotation_libraries)

bioinfo00/xnnet0 documentation built on Sept. 16, 2020, 8:22 a.m.