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 neuralnet trainPredict arguments - inputs fmla, data

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(neuralnet)
neuralnet.method <- c("slr", "sag", "rprop-", "rprop+", "backprop")
hyperParams.neuralnet <- function(method, ...) {
    return(list(iter=maxit1storderC, threshold=0.2, linear.output=TRUE))
}
NNtrain.neuralnet <- function(x, y, dataxy, formula, hidden_neur, method, hyperParams, ...) {
    
    hyper_params <- do.call(hyperParams, list(method, ...))
    
    if(method == "backprop") {learningrate <- 0.001} else {learningrate <- NULL}
    
    neuralnet::neuralnet(formula = formula, data = dataxy, hidden = hidden_neur, algorithm=method,
                        threshold=hyper_params$threshold, linear.output=hyper_params$linear.output,
                                    stepmax = hyper_params$iter, startweights = NULL, act.fct = "tanh",
                                    learningrate = learningrate)
}
NNpredict.neuralnet <- function(object, x, ...)
  as.numeric(predict(object, newdata = x))

NNclose.neuralnet <- function()
  if("package:neuralnet" %in% search())
    detach("package:neuralnet", unload=TRUE)
neuralnet.prepareZZ <- list(xdmv = "d", ydmv = "d", zdm = "d", scale = TRUE)

if(FALSE)
res <- train_and_predict_1data(1, neuralnet.method, "NNtrain.neuralnet", "hyperParams.neuralnet", "NNpredict.neuralnet", 
                               NNsummary, "NNclose.neuralnet", NA, neuralnet.prepareZZ, nrep=2,
                               echo=TRUE, doplot=FALSE, echoreport=0,
                               pkgname="neuralnet", pkgfun="neuralnet", rdafile=TRUE, odir=odir)

2 Launch package’s trainPredict

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

3 Results

#print(res)
kable(apply(res[,,1,], c(3,1), min), caption = "slr")%>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

slr
	RMSE	MSE	MAE	WAE	time
mDette	0.5081	0.2581	0.3846	2.0598	2.00
mFriedman	0.0782	0.0061	0.0684	0.1748	6.19
mIshigami	0.6889	0.4745	0.5116	2.9555	6.65
mRef153	3.5139	12.3475	2.5008	13.5814	0.00
uDmod1	0.1212	0.0147	0.0911	0.2957	0.03
uDmod2	0.1031	0.0106	0.0830	0.2699	0.01
uDreyfus1	0.2062	0.0425	0.1723	0.4822	0.00
uDreyfus2	0.1478	0.0218	0.1131	0.4772	0.00
uGauss1	2.6835	7.2009	2.1664	7.2250	0.12
uGauss2	3.6199	13.1039	2.6638	14.2330	0.06
uGauss3	3.4612	11.9802	2.7154	9.4444	0.03
uNeuroOne	0.2839	0.0806	0.2315	0.5901	0.00

kable(apply(res[,,2,], c(3,1), min), caption = "sag")%>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

sag
	RMSE	MSE	MAE	WAE	time
mDette	2.1873	4.7843	1.4869	14.3579	10.05
mFriedman	0.0810	0.0066	0.0731	0.1637	14.37
mIshigami	3.6898	13.6148	2.9776	13.1137	23.18
mRef153	3.5321	12.4754	2.4879	13.5064	0.09
uDmod1	0.0986	0.0097	0.0751	0.2445	0.09
uDmod2	0.0533	0.0028	0.0404	0.1336	0.17
uDreyfus1	0.0845	0.0071	0.0681	0.2469	0.00
uDreyfus2	0.1380	0.0191	0.1095	0.4437	0.00
uGauss1	2.3728	5.6303	1.8438	6.4944	0.95
uGauss2	3.5764	12.7904	2.6518	12.8941	0.31
uGauss3	2.6882	7.2265	2.1073	7.6228	0.03
uNeuroOne	0.2875	0.0826	0.2373	0.5672	0.01

kable(apply(res[,,3,], c(3,1), min), caption = "rprop-")%>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

rprop-
	RMSE	MSE	MAE	WAE	time
mDette	0.4446	0.1977	0.3353	1.8921	0.33
mFriedman	0.0079	0.0001	0.0062	0.0282	1.43
mIshigami	0.5951	0.3541	0.4313	2.5546	1.05
mRef153	3.4256	11.7344	2.4574	11.8535	0.00
uDmod1	0.1120	0.0125	0.0807	0.2938	0.01
uDmod2	0.0909	0.0083	0.0733	0.2001	0.02
uDreyfus1	0.2014	0.0405	0.1518	0.5512	0.00
uDreyfus2	0.1201	0.0144	0.0930	0.3439	0.00
uGauss1	2.5005	6.2525	1.9565	7.0535	0.09
uGauss2	3.3776	11.4084	2.6799	9.4165	0.05
uGauss3	2.7630	7.6341	2.1991	7.5066	0.01
uNeuroOne	0.2875	0.0826	0.2378	0.5537	0.00

kable(apply(res[,,4,], c(3,1), min), caption = "rprop+")%>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

rprop+
	RMSE	MSE	MAE	WAE	time
mDette	0.5886	0.3464	0.4360	2.5436	0.42
mFriedman	0.0088	0.0001	0.0071	0.0265	1.64
mIshigami	0.3844	0.1478	0.2943	1.6935	1.21
mRef153	3.2534	10.5849	2.3087	13.0795	0.00
uDmod1	0.0875	0.0076	0.0647	0.2547	0.02
uDmod2	0.0639	0.0041	0.0526	0.1434	0.02
uDreyfus1	0.1678	0.0282	0.1222	0.3999	0.00
uDreyfus2	0.1514	0.0229	0.1172	0.4654	0.00
uGauss1	2.5278	6.3895	1.9628	7.4449	0.07
uGauss2	3.6927	13.6357	2.7311	11.8958	0.05
uGauss3	2.5436	6.4699	2.0353	7.5269	0.03
uNeuroOne	0.2842	0.0808	0.2328	0.5227	0.00

kable(apply(res[,,5,], c(3,1), min), caption = "backprop")%>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"))

backprop
	RMSE	MSE	MAE	WAE	time
mDette	8.1656	66.6770	6.5262	36.2385	13.12
mFriedman	0.2348	0.0551	0.1880	0.6346	13.55
mIshigami	3.6898	13.6148	2.9776	13.1137	22.03
mRef153	3.9001	15.2109	2.8803	14.3666	0.04
uDmod1	0.1520	0.0231	0.1240	0.4189	0.20
uDmod2	0.1110	0.0123	0.0888	0.2368	0.25
uDreyfus1	0.2960	0.0876	0.2458	0.6437	0.01
uDreyfus2	0.2227	0.0496	0.1600	0.6753	0.02
uGauss1	2.6408	6.9736	2.0431	7.9787	0.39
uGauss2	3.2221	10.3822	2.5058	11.6109	0.61
uGauss3	3.3962	11.5339	2.6348	10.0509	0.23
uNeuroOne	0.2870	0.0824	0.2372	0.5080	0.13

NNbenchmark Generic Function with neuralnet

All members of GSoC 2019 & 2020

2020/08/07

1 Setup

1.1 Packages and options

1.2 Datasets to test

1.3 neuralnet trainPredict arguments - inputs fmla, data

2 Launch package’s trainPredict

3 Results