class_postprocess: Post-process classfication predictions

In mugiro/elm: Extreme Learning Machine for regression and classification

Description

Function to post-process the continuous prediction of a ELM in classfication

Usage

class_postprocess(object, ...)

Arguments

object	The instance to ELM class.
typePred	The type of post-processing implemented: "raw" No post-processing: ELM predictions. Continuos predictions that do not follow the definition of probability. "prob" Probability-like predictions. The softmax method is used for transformation. The predictions followed a probability function, as they are between 0 and 1 and sum 1. "label" Predicted labels from the probability-like continuous values.
ml_threshold	Threshold value to assign multi-label targets.

Details

In classification, labels for multi-class and multi-label cases are computed as follows. For multi-class, the output with higher continous value is selected. For multi-label, all outputs above ml_threshold are selected.

Value

The output matrix/vector with post-processed predtictors

mugiro/elm documentation built on May 23, 2019, 8:21 a.m.