1 Setup

1.1 Packages and options

library(NNbenchmark)
library(kableExtra)
options(scipen = 999)

1.2 Datasets to test

NNdataSummary(NNdatasets)
##      Dataset n_rows n_inputs n_neurons n_parameters
## 1     mDette    500        3         5           26
## 2  mFriedman    500        5         5           36
## 3  mIshigami    500        3        10           51
## 4    mRef153    153        5         3           22
## 5     uDmod1     51        1         6           19
## 6     uDmod2     51        1         5           16
## 7  uDreyfus1     51        1         3           10
## 8  uDreyfus2     51        1         3           10
## 9    uGauss1    250        1         5           16
## 10   uGauss2    250        1         4           13
## 11   uGauss3    250        1         4           13
## 12 uNeuroOne     51        1         2            7

1.3 deepdive trainPredict arguments - inputs x, y

if(dir.exists("D:/GSoC2020/Results/2020run03/"))
{  
  odir <- "D:/GSoC2020/Results/2020run03/"
}else if(dir.exists("~/Documents/recherche-enseignement/code/R/NNbenchmark-project/NNtempresult/"))
{  
  odir <- "~/Documents/recherche-enseignement/code/R/NNbenchmark-project/NNtempresult/"
}else
  odir <- "~"

nrep <- 10
maxit2ndorder  <-    200
maxit1storderA <-   1000
maxit1storderB <-  10000
maxit1storderC <- 100000

#library(deepdive)
deepdive.method <- c("gradientDescent","momentum","rmsProp","adam")
hyperParams.deepdive <- function(optim_method, ...) {
    
    if (optim_method == "gradientDescent")  {maxiter <- maxit1storderB; eta = 0.4} 
    if (optim_method == "momentum")         {maxiter <- maxit1storderB; eta = 0.8}
    if (optim_method == "rmsProp")          {maxiter <- maxit1storderA; eta = 0.8}
    if (optim_method == "adam")             {maxiter <- maxit1storderA; eta = 0.8}
  
    out <- list(iter = maxiter, method = optim_method, modelType = "regress", eta = eta, print = 1000)
    return (out)
}
NNtrain.deepdive <- function(x, y, dataxy, formula, neur, optim_method, hyperParams, NNfullformula, NNparam,...) {
    
    hyper_params <- do.call(hyperParams, list(optim_method, ...))
    
    NNreg <- deepdive::deepnet(x = x, y = y, c(neur), modelType = hyper_params$modelType,
                     iterations = hyper_params$iter, eta=hyper_params$eta, 
                     optimiser=hyper_params$method, printItrSize=hyper_params$print,
                     normalise = FALSE)
    
    return (NNreg)
}
NNpredict.deepdive <- function(object, x, ...)
  predict.deepnet(object,newData=x)$ypred
NNclose.deepdive <- function()
  if("package:deepdive" %in% search())
    detach("package:deepdive", unload=TRUE)
deepdive.prepareZZ <- list(xdmv = "d", ydmv = "d", zdm = "d", scale = TRUE)
if(FALSE)
res <- trainPredict_1data(1, deepdive.method, "NNtrain.deepdive", "hyperParams.deepdive", "NNpredict.deepdive", 
                               NNsummary, "NNclose.deepdive", NA, deepdive.prepareZZ, nrep=5, echo=TRUE, doplot=FALSE,
                               pkgname="deepdive", pkgfun="ann", csvfile=TRUE, rdafile=TRUE, odir=odir)

2 Launch package’s trainPredict

res <- trainPredict_1pkg(1:12, pkgname = "deepdive", pkgfun = "deepnet", deepdive.method,
  prepareZZ.arg = deepdive.prepareZZ, nrep = nrep, doplot = FALSE,
  csvfile = TRUE, rdafile = TRUE, odir = odir, echo = FALSE)

3 Results

kable(apply(res[,,1,], c(3,1), min), caption = "gradientDescent")%>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))
gradientDescent
RMSE MSE MAE WAE time
mDette 4.4310 19.6338 3.2628 20.7622 7.27
mFriedman 0.1474 0.0217 0.1154 0.4581 7.74
mIshigami 3.0218 9.1314 2.4940 10.2360 9.07
mRef153 4.1805 17.4764 3.1843 14.4946 5.91
uDmod1 0.3353 0.1124 0.2912 0.7067 5.33
uDmod2 0.2699 0.0729 0.2369 0.5509 5.30
uDreyfus1 0.3429 0.1176 0.2801 0.7346 5.22
uDreyfus2 0.3570 0.1275 0.2905 0.8478 5.19
uGauss1 23.6597 559.7792 19.1988 57.7174 6.05
uGauss2 28.8118 830.1176 25.1770 67.7823 5.95
uGauss3 32.2441 1039.6844 27.7063 70.1972 5.93
uNeuroOne 0.3666 0.1344 0.3105 0.6748 5.15 
kable(apply(res[,,2,], c(3,1), min), caption = "momentum")%>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))
momentum
RMSE MSE MAE WAE time
mDette 4.1990 17.6313 3.1011 18.5512 7.39
mFriedman 0.1363 0.0186 0.1061 0.4860 8.00
mIshigami 2.5791 6.6519 2.0107 8.7569 9.11
mRef153 3.8424 14.7643 2.7879 15.0474 6.11
uDmod1 0.3320 0.1103 0.2891 0.7441 5.57
uDmod2 0.2656 0.0705 0.2355 0.5269 5.48
uDreyfus1 0.3429 0.1176 0.2801 0.7341 5.49
uDreyfus2 0.3570 0.1275 0.2907 0.8468 5.39
uGauss1 16.2557 264.2462 13.0878 49.2409 6.23
uGauss2 28.8104 830.0390 25.1756 67.8020 6.17
uGauss3 32.2413 1039.4994 27.7055 70.2077 6.18
uNeuroOne 0.3544 0.1256 0.3001 0.6152 5.39 
kable(apply(res[,,3,], c(3,1), min), caption = "rmsProp")%>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))
rmsProp
RMSE MSE MAE WAE time
mDette 2.7205 7.4011 1.8705 16.1780 0.75
mFriedman 0.1287 0.0166 0.0990 0.4133 0.78
mIshigami 2.6728 7.1439 2.3060 7.1452 0.87
mRef153 4.0384 16.3091 3.0290 15.7099 0.62
uDmod1 0.1728 0.0299 0.1257 0.4478 0.56
uDmod2 0.2252 0.0507 0.1580 0.5513 0.55
uDreyfus1 0.1184 0.0140 0.0878 0.3446 0.54
uDreyfus2 0.1625 0.0264 0.1232 0.4839 0.54
uGauss1 23.5833 556.1721 19.1113 57.9958 0.64
uGauss2 18.3426 336.4509 12.6957 49.5614 0.61
uGauss3 20.7977 432.5437 15.4531 49.0396 0.62
uNeuroOne 0.3161 0.0999 0.2695 0.5981 0.55 
kable(apply(res[,,4,], c(3,1), min), caption = "adam")%>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))
adam
RMSE MSE MAE WAE time
mDette 3.0971 9.5922 2.0640 18.6373 0.74
mFriedman 0.0875 0.0077 0.0764 0.2699 0.78
mIshigami 2.5913 6.7150 2.0819 10.0604 0.88
mRef153 3.7605 14.1414 2.6651 15.3279 0.64
uDmod1 0.1178 0.0139 0.0797 0.4868 0.57
uDmod2 0.3189 0.1017 0.2113 0.7209 0.56
uDreyfus1 0.0304 0.0009 0.0265 0.0644 0.56
uDreyfus2 0.1149 0.0132 0.0907 0.2749 0.54
uGauss1 20.5179 420.9855 16.5474 47.6354 0.62
uGauss2 16.8843 285.0805 11.6361 46.6218 0.61
uGauss3 10.0466 100.9345 7.3484 31.6838 0.61
uNeuroOne 0.2946 0.0868 0.2459 0.5582 0.56