Benchmarking wrapper for darch

Description

Simple benchmarking function which wraps around the darch function for users who can't or don't want to use the caret package for benchmarking. This function requires the foreach package to work, and will perform parallel benchmarks if an appropriate backend was registered beforehand.

Usage

1
2
3
darchBench(..., bench.times = 1, bench.save = F,
  bench.dir = "./darch.benchmark", bench.continue = T, bench.delete = F,
  bench.seeds = NULL, output.capture = bench.save, logLevel = NULL)

Arguments

...

Parameters to the darch function

bench.times

How many benchmark runs to perform

bench.save

Whether to save benchmarking results to a directory

bench.dir

Path (relative or absolute) including directory where benchmark results are saved if bench.save is true

bench.continue

Whether the benchmark is to be continued from an earlier run. If TRUE, existing benchmark results are looked for in the directory given in bench.dir and new results are appended. If both this and bench.continue are FALSE and the directory given in bench.dir does already exist, the training will be aborted with an error.

bench.delete

Whether to delete the contents of bench.dir if bench.continue is FALSE. Caution: This will attempt to delete ALL files in the given directory, use at your own risk!

bench.seeds

Vector of seeds, one for each run. Will be passed to darch.

output.capture

Whether to capture R output in .Rout files in the given directory. This is the only way of gaining access to the R output since the foreach loop will not print anything to the console. Will be ignored if bench.save is FALSE.

logLevel

futile.logger log level. Uses the currently set log level by default, which is futile.logger::flog.info if it was not changed. Other available levels include, from least to most verbose: FATAL, ERROR, WARN, DEBUG, and TRACE.

Value

List of DArch instances; the results of each call to darch.

See Also

Other darch interface functions: darchTest, darch, plot.DArch, predict.DArch, print.DArch

Examples

1
2
3
4
5
6
7
8
## Not run: 
data(iris)
modelList <- darchBench(Species ~ ., iris, c(0, 50, 0),
 preProc.params = list(method = c("center", "scale")),
 darch.unitFunction = c("sigmoidUnit", "softmaxUnit"),
 darch.numEpochs = 30, bench.times = 10, bench.save = T)

## End(Not run)

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.