Training an MLP with a weighted set of trees

ROOT
1

Hello,

I have a list of text files that I want to use to create trees with different weights,
then combine these trees into one and use it to train a TMultiLayerPerceptron.

My goal, is for the TMultiLayerPerceptron to use during training the individual
weights per tree when trains itself.

I am thinking something on the lines:

a) Create from each TXT file a ROOT tree: tree_1->ReadFile(“file_1.txt”," … list of vars here …");
b) Set the weight for each individual tree: tree_1->SetWeight(wght_1);
c) Combine these trees into one tree “tree_all” How do I do that?
d) Feed the tree “tree_all” to my MLP

Any ideas on how to do achieve that?

Thanks in advance,
Stelios.

2

Hi,

You might get more information/answers by posting on the Math and Statistics forum rather than the general support.

Cheers,
Philippe.