Crashing when using pyroot and simple neural network in Keras via TMVA

TMVA
1

So I am using pyroot with the tmva feature through it to train a simple neural network constructed using KERAS (1 hidden layer with 3 neutron,5 input and 2 output). The training data is in .root with about 10^5 events.

However whenever I try to train it, my desktop freezes which I suspect that I may lack in memory. However this is weird as the number of weights for the network should not have used up so much memory given that it is just a simple neural network.

Has anyone faces the same problem when using tmva in pyroot to train a neural network from Keras?

2

Hi,

Could you provide the set-up (code + some input root file) so we can reproduce it?

Cheers,
Kim

3

Hey Kim,

I actually managed to found out the reason why its crashing, it seem that the tmva for it to run in my machine, I cannot let it handle more than ~10 variables/branches as anything more than it just straight up crashes(probably due to amount of events I have? ~10^5 and above). I am actually going to transition into purely python and just using root_pandas/root_numpy which depend on ROOT as there are some ML tools that I am thinking of testing it which are available mostly in python.

Thanks for the reply though

Danny