Discussion of article "Deep Neural Networks (Part VIII). Increasing the classification quality of bagging ensembles"

New comment
 

New article Deep Neural Networks (Part VIII). Increasing the classification quality of bagging ensembles has been published:

The article considers three methods which can be used to increase the classification quality of bagging ensembles, and their efficiency is estimated. The effects of optimization of the ELM neural network hyperparameters and postprocessing parameters are evaluated.

The figure below provides a simplified scheme of all calculations: it shows the stages, the used scripts and data structures.


Fig. 11. Structure and sequence of the main calculations in the article.

Author: Vladimir Perervenko

 

Discussion and questions on the code can be done in the branch

Good luck

 
After choosing the best 7 ensembles and classifying it[=1,0,1] ,I would like to extract the data to train it on a Keras model, but I can't seem to find the specific dataframes.
 
geraldkibz:
After choosing the best 7 ensembles and classifying it[=1,0,1] ,I would like to extract the data to train it on a Keras model, but I can't seem to find the specific dataframes.

Figure 11 shows the structural scheme of calculations. Above each stage is the name of the script. Under each stage is the name of the resulting data structure. What data do you want to use?

 

If you want to use the averaged continuous predictions of the seven best ensembles, then they are in the structure

testX1[[k]]$TrainYpred[ ,j]

  k = c(origin/repaired/removed/relabeled)

 j = c( half, mean, med, both)  

If you need the predictions of the seven best in binary form, then they are in the structure

VotAver[[k]]Train.clVoting[1001,j]
VotAver[[k]]Test.clVoting[501,j]
VotAver[[k]]Test1.clVoting[251,j]
 
Hi Vlad - any tips on how to connect the above NN to the EA that you had uploaded a few years ago? I am keen to test with a slightly different dataset
 

Hello again,

I am getting the following errors, which I cant resolve - any tips?


Error 1: "in { : task 1 failed - "object 'History' not found" , when I run the following code segment:

#---OptPar------
evalq({
  foreach(i = 1:4) %do% {
    OPT_Res[[i]] %$% History %>% dp$arrange(desc(Value)) %>% head(3)
  } -> best.res
  names(best.res) <- group
}, env)
evalq({
  foreach(i = 1:4) %do% {
    OPT_Res1[[i]] %$% History %>% dp$arrange(desc(Value)) %>% head(3)
  } -> best.res1
  names(best.res1) <- group
}, env)

Not sure where the History object is created, and couldnt find it at all in the github repo in the various .R files for this article


Error 2: "Yts" not found, when I run the following code segment:

#---test-aver--------
    foreach(i = 1:n, .packages = "elmNN", .combine = "+") %:%
      when(i %in% bestNN) %do% {
        predict(Ens[[i]], newdata = Xtest1[ , bestF])} %>%
      divide_by(length(bestNN)) -> ensPred
    th <- GetThreshold(ensPred, Yts$Ytest1, type[th2])
    ifelse(ensPred > th, 1, 0) -> ensPred
    Evaluate(actual = Ytest1, predicted = ensPred)$Metrics$F1 %>%
      mean() %>% round(3) -> Score

Also not sure when/how the "Yts" is being created


I think both of these might be resolved by a piece of code that may be missing from the github repo?

Would appreciate any help you can provide, many thanks in advance

New comment