pkg/R/m_ExponentialWeights.R

   1 #' @include b_LinearAlgorithm.R
   2
   3 #' @title Exponential Weights Algorithm
   4 #'
   5 #' @description Exponential Weights Algorithm.
   6 #' Inherits \code{\link{LinearAlgorithm}}
   7 #'
   8 #' @field alpha Importance of weights redistribution, in [0,1]. Default: 0
   9 #' @field grad Whether to use or not the (sub)gradient trick. Default: FALSE
  10 #'
  11 ExponentialWeights = setRefClass(
  12     Class = "ExponentialWeights",
  13
  14     fields = c(
  15         alpha = "numeric",
  16         grad = "logical"
  17     ),
  18
  19     contains = "LinearAlgorithm",
  20
  21     methods = list(
  22         initialize = function(...)
  23         {
  24             callSuper(...)
  25             if (length(alpha) == 0 || alpha < 0. || alpha > 1.)
  26                 alpha <<- 0. #no redistribution
  27             if (length(grad) == 0)
  28                 grad <<- FALSE
  29         },
  30         predict_noNA = function(XY, x)
  31         {
  32             K = ncol(XY) - 1
  33             if (K == 1)
  34             {
  35                 #shortcut: nothing to combine
  36                 finalWeight = 1.
  37             }
  38
  39             else
  40             {
  41                 X = XY[,names(XY) != "Measure"]
  42                 Y = XY[,"Measure"]
  43                 finalWeight = .C("ew_predict_noNA", X = as.double(t(X)), Y = as.double(Y), n = as.integer(nrow(XY)),
  44                     K = as.integer(K), alpha=as.double(alpha), grad = as.integer(grad), weight=double(K))$weight
  45             }
  46
  47             appendWeight(finalWeight)
  48             return (matricize(x) %*% finalWeight)
  49         }
  50     )
  51 )