mlampros / elmnnrcpp

Extreme Learning Machine using RcppArmadillo

elmNNRcpp ( Extreme Learning Machine )

The elmNNRcpp package is a reimplementation of elmNN using RcppArmadillo after the elmNN package was archived. Based on the documentation of the elmNN it consists of, "Training and predict functions for SLFN ( Single Hidden-layer Feedforward Neural Networks ) using the ELM algorithm. The ELM algorithm differs from the traditional gradient-based algorithms for very short training times ( it doesn't need any iterative tuning, this makes learning time very fast ) and there is no need to set any other parameters like learning rate, momentum, epochs, etc.". More details can be found in the package Documentation, Vignette and blog-post.

To install the package from CRAN use,

install.packages("elmNNRcpp")

and to download the latest version from Github use the install_github function of the devtools package,

remotes::install_github('mlampros/elmNNRcpp')

Use the following link to report bugs/issues,

https://github.com/mlampros/elmNNRcpp/issues

Citation:

If you use the code of this repository in your paper or research please cite both elmNNRcpp and the original articles / software https://CRAN.R-project.org/package=elmNNRcpp:

@Manual{,
  title = {{elmNNRcpp}: The Extreme Learning Machine Algorithm},
  author = {Lampros Mouselimis},
  year = {2022},
  note = {R package version 1.0.4},
  url = {https://CRAN.R-project.org/package=elmNNRcpp},
}

elmnnrcpp's Issues

OSELM or ORELM

Hi,

thank you for developing this amazing library.
I was wandering if you have any plan to implement OSELM or (even better) ORELM?

Best wishes

AttributeError: 'NoneType' object has no attribute 'format'

When I run the below command, I found this error

predictions= []
target= []
for i in range(numLags, len(df)-predictionStep-1):
  net.train(X[[i], :], T[[i], :])
  Y = net.predict(X[[i+1], :])
  predictions.append(Y[0][0])
  target.append(T[i][0])
print ("{:5}th timeStep -  target: {:8.4f}  |   prediction: {:8.4f} ").format(i, target[-1], predictions[-1])

gaussian weights not (0,1) doc issue

documentation issue: it says "normal_gaussian" weights are (0,1) but they are gaussian normal negative/positive

uniform_negative seed has no effect

If using "uniform_negative", the random seed has no effect so you always get the same weights if the input length and nhid values are the same. Looks like this is because you are using Mersenne twister random engine for those and seed isn't affecting it. (Maybe just rescale uniform to -1,+1?)